Global Burden of and Due to Occupational Factors 1

i

James Leigh,1 Petra Macaskill,1 Eeva Kuosma,2and John Mandryk1

We made estimates of absolute morbidity and mortality due of Disease project (1997). This was part of a larger study to occupational factors for the world using all available that estimated the total global disease burden. We present published data as of 1994, and, where no data were aggregate results and analyses by region and disease. We available, applying the most appropriate (in terms of similar estimate that approximately 100,000,000 occupational economy, race, and environment) (100,000 deaths) and 11,000,000 occupational age-Isex,ldiagnosis-specific incidence and mortality rates to (700,000 deaths) 'occur in the world each year. We known working population distributions. We report results regard these as very conservative estimates which, although according to economic groupings determined by the World unavoidably crude, can nevertheless provide a basis for Bank (World Development Report, 1993) and disease and priority planning at global level. (Epidemiology injury groupings according to The Global Burden 1999;10:626-631)

Keywords: global, , injury, incidence, mortality.

Occupational factors make an important contribution to the gle approach has been identified that can provide an global burden of disease. Work-related morbidity and accurate and comprehensive picture of the true extent of mortality not only results in suffering and hardship for the occupational disorders. Even within one country, the worker and his or her family, but also it adds to the overall different approaches used often lead to inconsistencies in cost to society through lost productivity and increased use of case definition that make it difficult to combine results. The medical and services. The cost to society has been differing thresholds of severity 'and biases associated with estimated at 2-14% of the gross national product in different various approaches can lead to marked variation in studies in different countries.1 estimates. It was the aim of this study to provide data to enable Methods for estimating rates of occupational disease and estimates of the overall global burden of occupational deaths from disease are particularly deficient, because many disease for inclusion in the overall Global Burden of of these diseases have multiple potential causes, including Disease study of Murray and Lopez.2 This required esti- life-style factors, and the long latency period that is mations in specified economic group, age, sex, and disease characteristic of many chronic diseases makes it difficult to categories. establish whether the condition is work related. This Estimation of the global burden of disease due to problem is further compounded by the general lack of occupational factors, including injury, is a difficult task. of primary health care providers in occupational Many advanced established market economies often have in medicine.3 place only fragmented reporting systems, and most less well Work-related traumatic fatalities and injuries or ill, developed countries usually have almost no reliable nesses of rapid onset are easier to identify than chronic information to assist in making attempts at estimation. diseases or deaths from disease. Although there is usually Developed countries use a wide range of data sources, little doubt about the link between occupational factors and including death records, hospital records, workers' com- an acute event, generally only the most severe and pensation claims, cancer registry records, significant injuries are logged and, hence, reported in records, surveys, and sentinel reports to estimate the extent official statistics. of work-related mortality and morbidity. No sin- Apart from inadequacies in primary reporting, collating, classifying, and publishing of data relating to occu- pationally caused disease and injury, there is the further factor of legal liability, which makes this area so difficult. From the 'National Institute of Occupational Health and Safety, From the outset there have been opposing forces at work Sydney, Australia, and 2Finnish Institute of Occupational Health, when employer responsibility for injury or disease is an Helsinki, Finland. issue. The history of occupational health has been a long Address correspondence to: James Leigh, National Occupational struggle by workers to establish attributability against Health and Safety Commission, GPO Box 58, Sydney 2001, NSW, employers who have sought to deny or reduce their liability. Australia. ©1999 by Epidemiology Resources Inc.

626 Epidemiology September 1999, Vol. 10 No. 5 GLOBAL OCCUPATIONAL DISEASE AND INJURY 627

Given the difficulties of estimating the true extent of In the special case of India only, because the ILO work-related morbidity and mortality, one approach that figures were quite implausibly low, we made an extrap- may provide a degree of consistency is to focus on olation on the basis of a l-year total surveillance program agreed-on compensable disorders. However, there are in one district.10 Nine contiguous villages (total almost certainly occupational exposures causing diseases population 24,000) in the Sonipat District of Haryana that have not yet been declared compensable.4 In addition, State were selected for complete surveillance of work. the reported compensated diseases in various countries related injury and disease over a 12-month period from will be a function of the political and legal systems of the July 1987 to July 1988. Trained field workers visited particular country. households every 2 weeks. In this period, 2,164 injuries Although many individual sample surveys for preva- were recorded. Of these, 576 were due to agricultural lence of various conditions in various subpopulations do implements, 295 to transport, 461 to children playing exist for some countries, no attempt will be made to with agricultural equipment, and 267 to household work. collate these systematically, as their use in global esti- Agricultural injuries constituted 14% of all ill-health mates would be problematic for the reasons outlined episodes. earlier. Similarly, there are a few studies estimating pop- Of the agricultural injuries, 72 (12.5%) were classed ulation attributable risk of various diseases due to occu- as moderate to severe, with an annual incidence rate of 3 pational factors. In particular, a review of occupational per 1,000 per year. attributable risk for cancer has been made by Simonato et It was estimated that in the states of Haryana, Punjab, al,' and the proportions reported there have been and Up, there are 5,000 to 10,000 deaths every year implicitly applied in some of the national estimates cited attributable to agriculture-related activities. Almost half in this paper. of these are caused by pesticide ingestion. It is estimated A detailed report of this work has been published as that 150,000 to 200,000 serious injuries occur. WHOIEHG/96.20.6 By extrapolation, assuming 70% of the Indian popu- lation (0.95 billion) works in agriculture, there are 16 million injuries per year, (2 million moderate to serious) Subjects and Methods and 53,000 deaths per year in agriculture alone. Another ESTIMATING RATES OF AND estimate of rates in India has been DISEASE made by Takala and Obadia11 based on Malaysian rates, We made a direct request in writing to countries to giving a figure of 36,500 deaths per year. We used a provide data on occupational morbidity and mortality. mean of 45,000 deaths per year for India and 17 million However, because only limited information was provided, injuries per year. two approaches, a "direct" and an “indirect”approach, were used to estimate the annual incidence of Occupational Disease occupational injury and disease in each of the eight World Reliable data on occupational disease are much more Bank regions, which are as follows: Established Market difficult to obtain than for injuries. Only the Scandina- Economies (EME), Former Socialist Economies of vian countries report occupational diseases incidence by Europe (FSE), India (IND), China (CHN), Other Asia and age and sex. Other nations provide more limited data, but Islands (OAI), Sub-Saharan Africa (SSA), Latin America relatively few use the World Health Organization- and the Caribbean (LAC), and Middle Eastern Crescent recommended categories. Some include occupational (MEC). These were used to satisfy the criteria of Murray disease in their injury statistics. and Lopez.2 For some countries, data for specific diseases attrib- uted to occupational factors by national reporting systems were available. We made no attempt to extrapolate DIRECT APPROACH disease-specific rates to countries without such data, as it Occupational Injuries and Fatalities was felt that local conditions would vary too much to The International Labour Office (ILO) report on global allow reliable estimates. However, national estimates of incidence of occupational injury7 was the major source of overall rates of occupational disease are probably more data on the number of injuries and deaths from injury per stable, and extrapolation at this broader level may provide year. The data are predominantly derived from corn, an indication of such rates for those countries for which pulsory -reporting systems or no data are available. compensation schemes. We used the most recent avail, By using the limited available data and applying rates able estimate for each country. We derived population from appropriate national or averaged data of countries sizes from data published by the ILO and the World or groups of countries closest in geography and economic Bank.8,9 For countries for which no injury data were development to those nations with no data, a complete available, we used the calculated injury (or mortality) rate global set of estimates by country can be obtained. In for a comparable country, or group of countries, to some cases, we give a range of estimates. We made no compute the expected number of injuries (or fatalities). attempt to allow for combined injury and disease We then combined these national estimates to obtain reporting. The inconsistencies that this introduces are estimates for each of the eight regions and summed probably not significant, given the overall degree of regional estimates to give worldwide figures. uncertainty associated with such data. 628 Leigh et al Epidemiology September 1999, Vol. 10 No. 5

TABLE 1. Occupational Nonfatal and Fatal Injuries injuries per year is about one-half of the estimates by (Direct Method) Mikheev' for injuries requiring greater than 3 days of lost time and for fatalities. The Mikheev estimates are ap- Injuries/Year Injuries/Year Region (Nonfatal) (Fatal) parently based on applying European Union injury rates to the world employed population using figures derived EME 9,481,000 14,100 FSE 3,000,000 10,300 from recent unpublished ILO data (M Mikheev, 1994, IND 17,000,000 45,000 personal communication), whereas the figures reported CHN 9,832,000 33,800 here derive from the ILO Yearbook of Labour Statistics OAI 2,384,000 11,400 7 SSA 770,000 9,900 1989-90, which are not greatly different from figures in LAC 2,077,000 8,900 the ILO Yearbook of Labour Statistics 1996. The likely MEC 1,134,000 8,440 explanation for this difference is underreporting. For WORLD 141,813 45,696,000 instance, the ILO-reported figures for India (200,000 EME = established market economies; FSE former Socialist economies of nonfatal injuries per year, 687 fatal injuries per year) are Europe; IND = India; CHN = China; 0A1 other Asia and Islands; SSA = Sub-Saharan Africa; LAC = Latin America and Caribbean; MEC = Middle so low that they are not believable. Even for developed Eastern Crescent. countries such as the United States, for which the figures do look plausible, it has been demonstrated that surveil- lance systems in place for estimating the number of work-related traumatic fatalities will significantly under, INDIRECT APPROACH estimate the true number of cases. 15,16 A 1995 report of Occupational Injury and the U.S. National Safety Council17 gives an even lower Disease figure for injuries (10,228,264), and this is also based on We made approximate estimates of occupational disease ILO data up to 1991. Takala,18 by using the best estimates and injury burden for the categories specified by Murray of fatal injury rates, including traffic at work, 2 and Lopez [specific pneumoconioses, chronic respiratory applied to working population data, estimates global disease (excluding specific pneumoconioses), mus- annual fatal work-related injuries as 346,552. This is culoskeletal conditions, accidental injury, cancer, neu- about three times higher than our estimate and 35% ropsychiatric disorders, pesticide poisoning, other higher than that of Mikheev.1 poisoning, skin conditions, and ,induced hearing In many instances the criteria for counting a work- loss by age, sex, and region] as follows. related injury are not clear. Not all countries count We derived the number of incident cases per year for lost-time injuries, and for those that do, the statistics may each disease category from the age-/sex-/disease-specific include any greater than 3-day and/or greater than 5-day incidence rates of occupational disease in Finland for 12 lost-time injuries. In the reported results, the choice of 1993. These data are the best defined, and the distri- rate for estimations was based on what seemed the most bution is similar to that of the other Scandinavian reliable reported data that were internally consistent and countries. Absolute numbers of cases in each disease plausible. category were estimated by applying the age-/sex-specific The criteria for work-related fatalities in particular rates to the appropriate age/sex population subgroups. have been recently defined and the difficulties discussed For EME and FSE, these rates were applied to the pop- in the U.S. and Australia.19 Such difficulties would be ulation distribution. For IND, CHN, OAI, SSA, LAC, and magnified considerably in less well developed countries. MEC, the same approach was used, but twice the Even among well developed countries, for example, dif- incidence rate for Finland was applied. For injury, age- 13 ferent nations deal differently with commuting accidents specific injury rates were taken from Australia, with a and specific industries. Some combine injuries and dis- doubling again applied for IND, CHN, OAI, SSA, LAC, ease for reporting purposes. The tendency to underreport and MEC. For China, a specific Chinese rate for pesticide 14 work-related injuries and traumatic fatalities is illustrated poisoning, not age/Sex specific, was applied. The age by the number reported for India. categories used were: 0-14, 15-44, 45-59, and 60+ years. Population totals for the eight World Bank regional Occupational Disease populations were based on World Bank data for 1990.8 The estimated number of work-related diseases (inci- dence and mortality) for each region is given in Table 2. We made estimates of age/sex distributions for World 6 Bank regions by applying distributions based on ILO (See also WHO 96.20, discussed above, for disease- data9 to the World Bank totals, as detailed in the World specific incidence per year and mortality per year for Health Organization report WHO/EHG/96.20.6 countries for which data were available. These countries were predominantly from the EME region). The results Results and Discussion confirm that reporting of occupational disease varies DIRECT APPROACH widely and that not all reported disease is compensated. Injury The estimated number of work-related injuries and injury INDIRECT APPROACH fatalities per year is presented for each of the regions in Occupational Disease and Injury 6 Table 1 (see report WHO/EHG/96.20 for details). The As described above, the Finnish distribution of rates for estimated number of nonfatal injuries and fatal 199312 was used as the best indicator of the distribution 1

1

Epidemiology September 1999, Vol. 10 No. 5 GLOBAL OCCUPATIONAL DISEASE AND INJURY 629

TABLE 2. Occupational Disease (Direct Method) TABLE 4. Estimated Annual Incidence of Occupational Injury Region Incidence/Year Mortality/Year and Disease in the World (Indirect Method)

Number of New Cases EME 837,400-895,500 109,800-113,400 Per Year FSE 201,000-206,700 46,800 IND 924,700-1,902,300 121,000 Injuries 100,688,000 CHN 88,300-2,537,900 8,600-161,500 Diseases 10,684X0 OAI 711,500-1,463,700 93,100 Pesticide poisoning 109,000 SSA 537,400-1,105,600 70,400 Other poisoning 122,000 LAC 407,400-803,000 64,200 Cancer 191,000 MEC 533,000-1,096,600 69,800 Mental disorders 318,000 WORLD 4,240,700-10,010,800 583,700-704,200 Pneumoconioses 453,000 Noise-induced 1,628,000 EME = established market Economies; FSE former Socialist economies of Skin disorders 1,895,000 Europe; IND = India; CHN = China; OAI other Asia and islands; SSA = Chronic 2,631,000 Sub-Saharan Aftica; LAC = Latin America and Caribbean; MEC = Middle Musculoskeletal disorders 3,337,000 Eastern Crescent. of disease in EME and FSE regions. The disease catego- other countries. However, the indirect method does ries used for the age-by-sex breakdowns did not corre- provide an estimate of how these cases are likely to be spond precisely to the required groupings, and hence the distributed by age and sex. These estimates are consid- rates by sex that were reported in the appropriate cate- erably lower than those of Mikheev1 (68-157 million new gories of disease were used in conjunction with these data cases per year). Mikheev used a correction coefficient of and the age-by-sex distribution of the Finnish pop- 8.7 times for estimating incidence in developing countries ulation to estimate the rates. The total population was from that of Nordic countries. The rationale for this was used as the denominator, because many occupational based on dividing the unweighted average incidence rates diseases have a long latency period and, hence, cases are of Sweden and Finland (regarded as high rates) by the i often diagnosed among persons who have retired and are unweighted average incidence rates of Poland and Russia no longer economically active. The rates used to corn- (regarded as low rates due to presumed underreporting, pute the indirect estimates for the EME and FSE regions poor occupational health and safety services, and are given in Table 3. These rates, doubled, were applied to malnutrition). We can see little justification for this and i population data for the remaining regions to estimate the actually thought the system of reporting in Poland quite expected number of cases of disease. good,20 using it as a basis for estimating other ESE A breakdown for injury and each disease category of numbers. In any case, there is extreme arbitrariness in any the estimated number of cases by region, age, and sex is such exercise, and the two different approaches at least provided in WHO/E1-1G/96.20.6 The overall numbers provide bounds covering the truth. estimated for injuries and each disease category are given Occupational injury incidence based on the indirect in Table 4. method is more than twice as high as the estimate based On the basis of this indirect approach, we would expect on the direct approach. Given that the indirect method approximately 10.7 million new cases of occupational assumed a threshold of at least 5 days lost time, the disease to be identified per year. This figure is consistent estimate of 100.7 million cases per year would represent with the 10 million cases estimated at the upper end of the more severe cases only, which in rum implies that the the range using the direct method. This level of agreement official national statistics published by many countries between the two methods is not surprising given that must be regarded as an underestimate of the actual disease data were not available for most countries and, number of work-related injury cases. The indirect esti- hence, the direct estimate was based largely on applying mate, based on Australian rates and a doubling for all disease rates for the EME region to

TABLE 3. Occupational Disease Rates Assumed in Calculations (Indirect Method)

Expected Number of Cases per Million Population Males Females Age 0-14 15-44 45-59 60+ 0-14 15-44 45-59 Pneumoconioses 0 86 246 295 0 0 72 54 Chronic Respiratory 4.1 172 1,843 1,775 2.1 225 480 54 disease Musculoskeletal 31 774 491 59 64 540 480 18 Cancer 0 4.3 147 295 0 0 12 9 Mental disorders 0 43 123 30 0 45 120 18 Pesticide poisoning 4.1 4.3 12 5.9 4.3 4.5 12 3.6 Other poisoning 4.1 21 49 15 4.3 9 24 9 Skin disorders 10 265 273 12 24 415 559 13 Noise-induced 0 177 1,889 556 0 8 127 31 hearing loss 630 Leigh et al

Sub-Saharan Africa China Latin America & Caribbean Formerly Socialist Economies of Europe India Other Asian and Islands Middle Eastern Crescent Established Market Economies OI: Occupational injuries OD: Occupational diseases WP: Working population FIGURE 1. Estimated number of new cases of occupational injuries and occupational diseases, per year, using the indirect method (all estimates in millions). countries other than EME, FSE, is comparable with that for injury, including sample survey data with large ad- of Mikheev,l based on European Union rates for injuries justments for underreporting. For injury deaths, adjust- requiring greater than 3 days of lost time. ments were also made. For disease incidence and mor- Figure 1 shows graphically occupational disease and tality, an attributable proportion method was used. Some injury (incidence) by the indirect approach and workforce of these were at the higher end of the generally accepted denominators for each of the eight regions. range. However, because of the weighting process in the calculation of DALYs, larger estimates, in morbidity in CONCLUSION particular, will not necessarily be reflected by Application of the results of this study to estimate dis- proportional increases in total number of DALYs. ability-adjusted life years (DALYs) resulted in about 38 We hope that this work will stimulate further refine- million DALYs, or 2.7% of all world DALYs being due ment of these methods and assessments. Local estimates, to occupational factors. In contrast, the highest risk factor particularly at the subnational and national levels, would studied was malnutrition, with 15.9% of all DALYs, be a valuable contribution to the further development of followed by poor water supply, sanitation, and personal this process. We would welcome the results of such local and domestic hygiene, with 6.8%; unsafe sex, with 3.5%; estimates for inclusion on future revisions of this study. alcohol, with 3.5%; tobacco, with 2.6%; hypertension, with 1.4%; physical inactivity, with 1.0%; illicit drugs, with 0.6%; and air pollution, with 0.5%. In comparing Acknowledgment health outcomes, road traffic accidents are responsible We thank Carlos Corvalán for his contributions to earlier versions of this paper for 6.5 DALYs per 1,000 persons globally, and and the report WHO/11HG/96.20, from which this material has been adapted. occupational risk factors are responsible for 18.6 DALYs per 1,000 persons in the working population.2 The numbers above are likely to produce a reaction. References For example, other estimates for occupational risk fac- 1 . Mikheev M. New epidemics: the challenge for international health work. In: tors indicate that morbidity and mortality numbers are New Epidemics in Occupational Health. Helsinki: Finnish Institute of underestimated in this report. A recent study, estimating Occupational Health, 1994;27-33. the numbers of occupational diseases and injuries in the 2. Murray QL, Lopez AD, eds. The Global Burden of Disease. vol 1. Cam- bridge, MA: Harvard School of , 1997. United States gives much higher estimates than those 3. Landrigan PJ, Baker DB. The recognition and control of occupational 16 given above. This study used a variety of data sources disease. JAMA 1992;266:676-680. Epidemiology September 1999, Vol. 10 No. 5 GLOBAL OCCUPATIONAL DISEASE AND INJURY 631

4 Rantanen J. New epidemics: questions to be addressed. In: New Epidemics Australia, 1992-3. Canberra, Australian Government Publishing Service, in Occupational Health. Helsinki: Finnish Institute of Occupational Health, 1995. 1994;5-16. 14. Chen S-Y. Incidence of occupational disease in China. Dis Monitoring 5. Simonato L, Vineis P, Fletcher AC. Estimates of the proportion of 1993;8:181-183. cancer attributable to occupational exposure. Carcinogenesis, 15. Suruda A, Emmett EA. Counting recognized occupational deaths in the 1988;9:11591165. 6. Leigh J, Macaskill P, Corvalan C, Kuosma E, Mandryk J. Global Burden United States. J Occup Med 1988;30:868-872. of Disease and Injury Due to Occupational Factors. WHO/E1-1G/96.20. 16. Leigh JP, Markowitz SB, Fahs M, Shin C, Landrigan PJ. Occupational Geneva: World Health Organization, 1996. injury and illness in the United States. Arch Int Med 199707:1557-1568. 7. International Labour Office (ILO). Occupational injuries. In: ILO 17. National Safety Council (USA). International Accident Facts. Itasca, IL: Yearbook of Labour Statistics 1990. vol. 19. Geneva: ILO, 1990;89 -90. National Safety Council, 1995. 18. Takala J. Fatal occupational accidents including traffic accidents at work 8. World Development Report. : World Bank, 1993. i 9. International Labour Office (ILO). Total and economically active popula [Paper presented to the Consultation on Methods for Health Impact tion. In: ILO Yearbook of Labour Statistics 1989-90. Geneva: ILO, Assessmerit in Environment and Occupational Health, World Health 1990. Organization/International Labour Office (VMO/ILO), Geneva, July 10. Mohan D, Patel R. Design of safer agricultural equipment: application of 19971. Geneva: WHO/II.0, 1997. ergonomics and epidemiology. Int J Ind Erg 1992; 10:301-309. 19. Stout N, Frommer MS, Harrison J. Comparison of work-related fatality 11. Takala J, Obadia I. International dimension of occupational and environ- surveillance in the USA and Australia. J Occup Accidents 1990;13:195- mental health. Environ Man Health 1997;8:197-198. 211. i 12. Finnish Institute of Occupational Health. Occupational diseases in Finland 20. S~kt Z, hydd J. Occupational diseases in Poland during the years 1984- in 1993. Helsinki: Finnish Institute of Occupational Health, 1994. 1992. Pol J Occup Med Environ Health 1993;6:299-308. 13. Worksafe Australia. Compendium of Workers Compensation Statistics,

t

1