Article Text
PDF
Abstract
Background Health and injury surveillance data of the highest achievable quality are needed in order to appropriately allocate scarce resources at the local and national levels.
Methods This is the first reported surveillance study of injury using a complete community sample in Viet Nam. Workplaces in Xuan Tien Commune most likely to benefit from intervention were identified and ranked by the magnitude of the problem (or highest injury count), the risk (highest incidence rates) and the burden (the effect of injuries on the livelihoods of workers).
Results 591 injuries occurring in the month prior to survey administration were recalled, which satisfied the injury case criteria of this study (the annualised incidence rate (IR) was 681 per 1000 residents). 482 were attributed to work activities (82%), yielding an annualised IR of 1001/1000 full time employee equivalents (FTE). The highest number of injuries occurred in the manufacturing sector (n=299), followed by agriculture with far fewer injuries (n=70). The highest rate of injury was in the transport, storage and communications sector (annualised IR 1583/1000 FTE), followed by manufacturing (1235/1000 FTE) and agriculture (844/1000 FTE).
Conclusion This study identified patterns of risk which, because data collection reflected work culture, are believed to be more reliable than those from previous studies. Interventions in the manufacture of machinery and equipment sector (the largest industry in the commune) would have the most impact in reducing occupational injuries. Despite the trend towards manufacturing, agriculture is still a high priority with a continuing substantial impact.
- Incidence
- injury
- surveillance
- Vietnam
- work-related
Statistics from Altmetric.com
Introduction
Viet Nam, historically an agrarian country, has recently seen gross domestic product (GDP) growth rates of 7–9% per year. The poverty rate, although still substantial, dropped 30% in less than 10 years (58.1% in 1993 vs 28.9% in 2002)1 demonstrating the positive affect of “Doi Moi”, the liberalisation of the economy in 1986.
During periods of rapid development, the health impacts of economic growth and the associated societal transformations are often not taken into consideration in policy discussions. While the effect of economic development on the incidence of injury can be substantial, it is often hidden by lack of data documenting the magnitude of the burden in economic and social costs. Viet Nam exemplifies this with its rapid development coupled with the absence of an effective injury registration system.
The objective of the current study was to quantify the burden of work-related injury in a single commune in Viet Nam to better understand the occurrence of work-related injury in a rapidly industrialising, agrarian society. During rapid industrialisation there is vagueness between work and non-work time, since “leisure time” is often not considered in the same way in an agrarian society as in a developed economy. From a Western perspective, work and home life in Viet Nam are very closely intertwined.2
Fila Bavi, a demographic surveillance site and epidemiological field laboratory, was established in Viet Nam in 1996 to facilitate the collection of appropriate health data, establish baseline health indicators and monitor change at the community level for policy makers and preventive health centres.3 Hang et al4 5 administered four cross-sectional household surveys during 2000 at 3-month intervals to collect information from adults sampled on injuries in the previous 3 months. The overall annual incidence of injury was reported as 89/1000 person-years. This rate was 30% lower than that of the USA (129/1000 persons) as reported by the National Health Interview Survey.6 The incidence rate where the circumstance of injury was “work-related” (including that occurring in the home, at school, due to road traffic, etc) was 17 per 1000 person-years.
In 2005, we developed an active injury surveillance study in Xuan Tien Commune in Viet Nam. Results on the demographics of work and work culture from a household survey administered in this community were reported previously.2 The principal findings were:
The overall injury incidence rate was 681 per 1000 residents in Xuan Tien, which is five to ten times higher than described in previous studies in Viet Nam.
Home and work activities were closely intermingled; over 28% of households had family-owned businesses.
Many workers work more than one job and, for those who do, one job is commonly in agriculture and the second in the manufacturing or service industries.
Despite intense development of the manufacturing and trade sectors, over 60% of land consists of rice paddies, and work and community life still revolve around the planting, growing and harvesting of rice.
The objectives of this paper are to quantify the burden of work-related injury, to identify high-risk conditions and environments, and to inform the choice of interventions to prevent future injuries. Workplaces in most need of intervention were identified by ranking the magnitude of the problem (or highest number of injuries), the risk (or highest incidence rates) and the burden (or impact of injuries on lost working time).
Methods
The study site
Xuan Tien Commune, Nam Dinh Province, is located 130 km southeast of Hanoi. With 2647 households and approximately 10 700 inhabitants, Xuan Tien was small enough for a 100% sample and yet large enough to have both a central agricultural component and a substantial industrial base with some medium and large enterprises (eg, textile and machinery manufacturing for export) and many family-owned businesses. Many communes throughout Viet Nam are in a similar stage of development.2
Survey administration
Twenty one volunteers (mainly health volunteers) administered a cross-sectional survey gathering information on demographics and self-reported injury to the 2647 households between April and December 2005. Health volunteers are either doctor's assistants or nurses stationed within the village or commune health station to offer health advice or treatment to people within their community. Volunteers received a 2-day training session on human subject protection and administration of the questionnaire.
Household survey design
The survey was developed in English and then, using a de-centring translation technique that gives equal weight to both languages, was translated into Vietnamese and then back into English to confirm both accuracy and cultural appropriateness.7
The household questionnaire consisted of three parts. The first included general information about each household, such as address, and the characteristics of family-owned businesses. The second included detail on demographics and all work performed by each family member.
In the third section of the survey, the head of household was asked about “hurts” to all family members regardless of age. We used the Vietnamese equivalent of the English term “hurt” instead of “injury” because it covered a broader, more comprehensive set of conditions than the Vietnamese word for “injury”.2 Through many discussions with our Vietnamese colleagues and pilot testing, we determined that asking about “đau” (“hurts”) was more likely to capture over-exertion events leading to injury, which we wanted to include in our estimates. We believe that we captured (after classification and review) what is termed “injury” in the West and, consequently, the term “injury” is used to describe our results.
We focused on injuries in the previous month, considering that this period provided for adequate recall and yet was long enough that there were likely to be a substantial number of events.2
In completing the third section, the volunteer provided the head of household with a printed list of “potential hurts” as examples of what should be reported. The head of household was asked, for each household member: “Was hurt in the past month enough to need care or to disrupt their normal activities for at least one day or longer?”. For those reporting an injury, more detail on the injury was obtained using a separate “hurt form”. If the individual was aged 18 or older, he/she was asked to respond with details of the incident. For younger family members the head of household completed the “hurt form”.
Assignment of work-relatedness
The main objectives for the follow-up on injuries were: (1) to gather information to assess work-relatedness, (2) to determine the events leading to injury, by asking narrative questions about the type of injury and the activities the person was involved in at the time of the injury and (3) to determine what happened after the injury. An independent reviewer assigned work-relatedness by using all the supporting information (eg, injury narrative, job title, etc). The judgements of the independent reviewer were compared with the answers to directed questions on work-relatedness: (1) “where were you hurt?” (public transportation, residence, workplace, public place, other), (2) “were you working when you were hurt?” and (3) “were you going to work or on the way home from work?”.
Severity measures
Self-reported severity had three components: (1) type of injury, (2) if and for how long the individual had to stop normal daily activities and (3) place of treatment (hospital, health station, traditional healer, self-treatment or other). Descriptions of the hurt incident were subsequently classified by a certified ICD coder into nature of injury according to the International Classification of Diseases (ICD-10) and into body part classification according to the International Labor Organization (ILO).8
Inclusion/exclusion criteria
Included were all cases that were likely due to an external cause, including all trauma, poisonings and musculoskeletal injuries. All ICD-10 classifications beginning with an “S” or a “T” classified as an injury, poisoning or certain other consequences of external causes, were included. This is a broader case definition than used by many occupational injury studies because it includes non-traumatic conditions such as sciatica as long as there was reasonable evidence of a trigger event (such as carrying, lifting, pouring, pushing, etc). Cases which were assigned an ICD-10 diagnosis code of “M” or “R” (mainly musculoskeletal conditions) were reviewed by two independent researchers to determine if they qualified as injuries using the following case information: description of the activities (including work) at the time of the injury, events after the injury, and age. The main criteria for exclusion were if no external attribution could be determined from the narrative or if the case indicated that the pain had persisted for a long period of time. For those cases where the reviewers disagreed (about 13% of “M” and “R” cases), the reviewers conferred and reached a consensus.
Work hours
The time spent the previous year working in each job was determined based on reported number of hours worked per day (appendix A describes the questions used concerning type of work and hours). For each worker the total hours worked per year were computed by summing the work hours for each job. However, 2% of the responses were missing a component necessary for this calculation. Missing information was imputed using the mean hours worked in each industry.
Industry classification
Job titles and main task descriptions and the description of the circumstance of the injury were used to classify the industry of every job reported and every work-related injury according to a modified ILO industry classification scheme (appendix B).9
Data analysis
Information from the household survey was analysed using Statistical Analysis Software (SAS) v 9.0. Incidence rates were computed based on the number of Xuan Tien residents (for total injuries) and number of full time employee equivalents (FTE) for work-related injuries. One FTE was assumed to be equivalent to 2000 work hours over a 1-year period. Annualised incidence rates are reported based on injuries in the last month multiplied by 12.
Industries were compared by: (1) injuries reported in the last month, (2) injury rates and (3) injury severity. Severity was assessed in two ways: from self-report of days lost from normal activities and from treatment venue. Since severity in days lost will depend on how long after the injury the case was interviewed, the overall assessment of days lost is conservative. The mean time between interview and injury was 20 days. Censoring occurred in 16 cases where the person had not returned back to normal duties by the time they were interviewed, but these cases did not appear to introduce bias into any of the comparisons reported here (data not shown).
Human subject protection
The household survey was submitted to and approved by the institutional review boards at the University of Massachusetts Lowell, the Liberty Mutual Research Institute for Safety, and the National Institute for Occupational and Environmental Health.
Results
Survey participation
In total, 2615 households completed the survey. Only 32 did not respond (response rate=99%). Our study population included 10 416 residents of Xuan Tien with an age and gender distribution similar to that found throughout Viet Nam.2 Overall, 5486 workers reported working 8478 jobs over a 1-year period or 5780 FTE.
Annual incidence rates of injury in Xuan Tien
A total of 601 injuries from all causes were reported in the month prior to survey administration. As 10 cases did not satisfy the case definition they were excluded, leaving 591 qualifying injuries. Calculated as the rate of injury per population, the incidence rate at 681 per 1000 residents was considerably higher2 than reported from Fila Bavi (89/1000 person-years4 5 10) and from a national survey (49/1000 residents11).
Work-related incidence of injury
Overall, 482 of the injuries reported were attributed to work activities (82%) by independent review, yielding an annualised work-related incidence rate of 1001/1000 FTE (table 1). Of these, 2.9% (n=17) occurred commuting to or from work. There was very good agreement concerning work-relatedness between the respondents' answers on the two directed questions on whether their injuries occurred while working or commuting and the independent reviewers' decisions based on injury narrative information. The directed questions were found to be both sensitive (97.5%) and specific (86.2%).
Self-reported incidence of injury
There was poor agreement, however, when using the self-identified place of injury to determine the proportion of injuries that were work related, as only 50% of those injured indicated a workplace as the location of their injury. Another 33% indicated their residence, despite clear narrative evidence that the injury was work related, confirming the ambiguity in Viet Nam when categorising the place of work as either workplace or home. The other possible locations were much less common—public transportation (8%) and a public place (4%).
The incidence rate of injury for non-workers (161/1000 non-working residents) was much lower than for workers (1149/1000 working residents). Non-workers were those residents who did not report any work in the last year. The head of household was asked for a reason if no work was indicated for a family member. There were 4931 such residents (47%) who did not report at least one job, the reason given for the large majority (82%) being that they were too young to work (or were in school). The remaining non-workers were either too old or retired, could not work for health reasons, were looking for a job, were soldiers, or were taking care of the family or home.
Several types of injuries were attributed exclusively to work activities (table 2). Crushing injuries, amputations and foreign body injuries were entirely associated with work. More than 80% of burns to the face, neck and eyes, open wounds to an upper or lower extremity, upper extremity contusions, and sprains and strains were reported as work related. Finally, 88% of all fractures of the skull were due to work.
All injuries and work-related injuries reported in the month prior to survey by nature of injury (n>9)
Work-related injury severity
Of those injured, 89% had to stop normal activities, with 62% stopping for a day or more and 15% for more than 5 days (table 3.) The mean lost time for all cases unable to resume normal activities was 3.5 days. By venue, 66 injuries (11%) were treated at a hospital, 142 (24%) at the commune health station, 17 (3%) by a private physician and 21 (4%) by a traditional healer, while 299 (51%) reported self-treating their injuries by using supplies at home or buying supplies at the local pharmacy.
Indicators of severity for all work-related injuries reported in the month prior to survey
Industry comparisons
Rank by magnitude
The highest number of injuries was in the manufacturing sector (n=299), while agriculture was second (n=70) and wholesale and retail trade third (n=52) (table 4). The manufacture of machinery and equipment, (predominantly the production of cement mixers and rice threshing machines) had the highest number of cases (n=167) within manufacturing, with foundries ranked second (n=48) and the manufacture of food products third (n=26).
Incidence of work-related injury* by industry (nFTE>75)
Rank by incidence rate
The highest injury rate was in the transport, storage and communication sector (1583/1000 FTE), but because there were only 11 injuries in this area, the confidence interval for the rate was quite wide, and chance may explain this rank. Manufacturing, with the second highest injury rate (1235/1000 FTE), was 35 times larger in terms of FTE. Agriculture had the third highest injury rate (844/1000 FTE).
Although manufacturing was clearly a high-risk activity, there was a wide range of risk among different types of manufacturing. Foundries had the highest incidence rate (2621/1000 FTE) and the manufacture of machinery and equipment had the second highest rate (1638/1000 FTE). The manufacture of bamboo products (1318/1000 FTE) ranked third. Comparatively low risks were seen in the manufacture of textiles and manufacture of food products and beverages (215/1000 and 495/1000 FTE, respectively; table 4). In agriculture, field work was more dangerous than breeding or butchering animals (998 vs 651/1000 FTE).
Rank by severity
In addition to high injury rates, the manufacturing sector (specifically, the manufacture of machinery) was responsible for all the amputations and crushing injuries in the commune. The severity of injuries in the transport, storage and communication sector was quite high (on average 4.2 days of normal activities were lost per case, and 73% of such injuries were formally treated at a hospital or commune health station). Although the injury rate in the manufacture of rubber and plastic products was ranked fifth out of eight for the manufacturing sector, the mean lost time per case was high at 11.3 days (table 4).
Work in agriculture in Xuan Tien: the effect on risk in manufacturing
We explored the impact of agricultural work since over 60% of workers did some work in the fields. We found that being employed in agriculture at any time over the year had an impact on injury rates. When isolating workers primarily engaged in manufacturing, those who spent at least some time in agriculture had higher injury rates overall (1256/1000 FTE) than those working solely in manufacturing (1141/1000 FTE). Furthermore, the time spent in their manufacturing work appeared to be riskier for those who also worked in agriculture (1333 injuries in manufacturing/1000 FTE worked in manufacturing) compared to those who worked only in manufacturing (1141 injuries in manufacturing/1000 FTE worked in manufacturing).
Discussion
This is the first reported surveillance study using a complete community survey in Viet Nam. More than 100 individuals from Viet Nam assisted in conducting this research, resulting in an unusually comprehensive capture of occupational injuries. The household survey was administered in less than 6 months by 21 volunteers to over 2500 households and resulted in a 99% response rate.
Burden of work
The questions used to assess work-relatedness were found to be both sensitive (97.5%) and specific (86.2%). In total, 82% of the injuries to all residents reported in this comprehensive survey were attributable to work. When analysis is restricted to the working age population (15–65 years of age), the work-related fraction rises to 91%. This, and the surprisingly high work-related injury rate (1001/1000 residents), reveal the considerable health burden from occupational injury in this community.
Although 82% of the injuries were attributed to work, only 50% of the cases identified the “workplace” as the place of injury. An additional 33% of cases reported that the place of injury was at “home” and it was determined that 68% of these injuries involved work at home. In rural Viet Nam, cooking areas as well as living areas can be in close proximity to equipment and materials used in the family business (including machine tools, power saws and foundry furnaces). Rarely are there physical barriers separating work environments from living areas, and children often play in the area where work is performed.
The only prior study in Viet Nam which considered work injuries, reported an overall nationwide work-related injury incidence rate of 7.0 per 1000 residents, and in the northern province 13.7 per 1000 residents.12 The authors of the study acknowledged, however, that the survey did not focus on work-related injuries and the information gathered was limited. Also, they did not provide statistics on how the findings on work related to the entire burden of injury. Consequently, we are unable to compare our current estimates to that study in detail, as little information on recall period, the sample, or the case definition for work-relatedness was reported.
Another issue which limits comparison of these results with prior studies is that we chose to include in our case definition any musculoskeletal conditions due to an external cause. This might, therefore, include some non-traumatic conditions such as sciatica which other studies would not consider “injuries”.
Recall bias is always a concern for cross-sectional injury surveys.4 10 13 14 That is why a priori we believed that a 1-month period was appropriate for this study, allowing for good recall, yet yielding sufficient numbers for analysis of risk patterns. While focussing on injuries in the last month may improve recall, this might also introduce a bias if the month was not representative, or if injuries, which actually occurred earlier, were incorrectly assigned to the most recent month. Most of the interviews were carried out 1–3 months after the June harvest, a busy time of year. Annualising from this period may have led to an over-estimation of the true annual average rates in agriculture and an underestimate of the rates in manufacturing if the pattern of injuries during the harvest affected reporting in the survey. However, the resulting injury rate in manufacturing was considerably higher than the rate in agriculture, making it possible that the manufacturing injury rate at other times of year could be even higher.
Although most of the land area in this community is given over to rice production, Xuan Tien has prospered in recent years because of its many rapidly expanding manufacturing industries. In addition, 28% of households have a family-owned business and many people work more than one job throughout the year.2 Many residents' wages doubled between 2005 and 2006 because of this intense economic activity, but the higher injury risk in manufacturing demonstrates the clear health cost of this development. These findings help identify industries where interventions would likely be most effective at reducing the burden of injuries. For example, it was clear that in Xuan Tien, machine guarding and housekeeping should be improved in the manufacture of machinery sector (which had the highest counts and second highest rates). In addition to high injury rates overall, this industry was responsible for all the amputations and crushing injuries in the commune. It is reasonable to conclude from these statistics that the negative health impact of this industry is substantial; the manufacture of machinery should be considered the highest priority industry for improvement.
The finding that workers who divide their time between agriculture and manufacturing seemed to have higher rates of injury during their manufacturing activities compared with those who only work in manufacturing was unexpected. The life of this community, like many others in Viet Nam, continues to revolve around the cycle of planting and harvesting of rice. The rice harvest is a community effort and many workers leave their factory jobs to help out in the fields at that time. One might hypothesise that fatigue during the harvest is a contributing cause, but we have no direct way to evaluate this. Careful study, perhaps with work diaries, of these workers switching back and forth between the fields and the workshop would be useful to try to identify ways to reduce these high risks. Despite the trend towards manufacturing, agriculture is still a high priority and its impact on injuries will likely be substantial for many years.
Because we observed an apparent modifying effect of agricultural work on risk of injury, and because agricultural work in Viet Nam has seasonal patterns, it is recommended that in the future, data on hours of work are collected through a census that is distributed across the entire year. In Xuan Tien for example, a random twelfth of the 10 416 commune members could be interviewed each month. Additionally, questions on work should identify the specific months as well as the number of months that each type of work is performed.
This study was able to identify these patterns of risk because of the comprehensive data that were gathered on hours of work. In addition, had we used only the number of workers whose primary job was in an industry (or agriculture) as the denominator for risk calculations, the patterns by industry would have been quite different. For example, the risk of injury in agriculture more than doubled when the rates per 1000 FTE were compared instead of simply the number employed.
Conclusion
A complete community survey administered in Xuan Tien commune in rural northern Viet Nam collected detailed data on the patterns of work and injury in a period when economic development is rapidly transforming an agrarian society into an industrial one. The commune's overall injury rate was very high at 681/1000 residents and 82% of all injuries were attributed to work. Because work and home life are closely intertwined, the location of many of these injuries was the home, but it was clear that they were actually attributable to paid work. Many workers (40%) work part of the year in the fields and part of the year in manufacturing and other industrial activities, often in small family-owned businesses. We were able to determine workplaces in Xuan Tien that would most likely benefit from intervention, and identified them by rank according to the absolute numbers of injuries, by risks and by injury severity.
This study greatly benefited from strong support within the community as well as from different levels of the national healthcare system, including local health volunteers, the commune health clinic, and provincial and national preventive health centres. The cooperation and trust of the community, government and healthcare system may partially explain the fact that this study found much higher estimates of injury incidence than had been reported previously.
What this paper adds
In developing economies such as Viet Nam, the effect of economic development on the incidence of injury can be hidden by a lack of data.
The current study sought to quantify the burden of work-related injury in a single commune in Viet Nam to better understand the occurrence of work-related injury in a rapidly industrialising, agrarian society.
We determined over 80% of the injuries in the community of Xuan Tien, reported from a cross-sectional survey, were attributable to work, with the greatest number of work-related injuries reported in manufacturing followed by agriculture with far fewer; however, there was a substantial modifying effect on the risk of injury in other industries for those also working in agriculture.
In Viet Nam, the workplace and home are closely intertwined and, while the location of many of the injuries was the home, these injuries arose from paid work.
Acknowledgments
The authors would like to thank Dr Ngo Quynh Sang who was the main physician and coordinator for health issues in the commune and completely dedicated to our research goals. None of this work would have been possible without him and we are greatly indebted to him. We would also like to express our deep thanks to Ms Tran Hai Yen, initially the translator on the project. However, she became much more than a translator and was instrumental in bridging the gaps in understanding between the two cultures. We also would like to express deep gratitude to the health volunteers of Xuan Tien who collected the surveys, interacted with household members, achieved a very high response and collected excellent information. In addition we would like to thank the dedicated staff of the psychophysical department of the National Institute for Occupational and Environmental Health for the many hours of organising files, entering and cleaning data as well as translating all questionnaires into English, and the party leaders who supported our work as well as health staff from the provincial medical centre and district health centre. Finally, we would like to thank Ms An Thanh Do, Ms Trang Dao and Ms Ginny Briggs for their important input during the initial study design, Valerie Newcomb for her hard work and creative input during classification of the data, and Patti Boelsen and Peg Rothwell for editorial input to the final manuscript.
Appendix A Survey questions on work (final back translation)
Work outside the home for pay (Question 10):
I would like to ask each person living in the household a number of questions about the work he or she has done outside [of the house] during the last year. That would be paid or unpaid work. I would like to start with you, then will ask each person in your family the same questions. If the work outside the home is not for someone else but for themselves or their family, please mark “x” at “self employed” in Q10b. Please list all the jobs, if one person does more than 1 job, please list the next job in the next line.
10a. Are you currently or were you last year working outside your home?
10b. For whom and where do you work? (self-employed please check box)
10c. Occupation and current work tasks?
10d. How much per month did you earn from each job?
10e. For how many months of last year did you do each job?
10f. On average, how many hours per day did you do each job?
10g. On average, how many days per week did you do each job?
Work inside the home for pay (Question 11):
To each household member whom I have listed, I would like to ask a number of questions on essential jobs they performed during the last year at home to create income. I would like to start with you and then with other family members with the same questions. Please list all the jobs. If one person does multiple jobs, please list the next job in the next line.
Questions 11b-g same as 10b-g listed above.
Work at home without pay (Question 12):
To each household member I would like to ask a number of questions on typical working activities such as husbandry, processing, and serving workers that they performed at home last year relating to the family business without receiving pay. I will start with you and then will ask each family member the same questions. Please list all the jobs. If one person does multiple jobs, please list the next job in the next line.
Questions 12b-g same as 12b-g listed above.
Why not working (Question 13):
Interviewer: Ask household members who answer no to all questions 10–12.
Why don't you work at home or in another place?
Appendix B Modified ILO industry list: classification of economic activities9
Footnotes
Work was organised in Viet Nam by the National Institute for Occupational and Environmental Health.
Funding The Liberty Mutual Research Institute for Safety provided funding for this study.
Competing interests None.
Ethics approval This study was conducted with the approval of the University of Massachusetts Lowell, the Liberty Mutual Research Institute for Safety and the National Institute for Occupational and Environmental Health.
Provenance and peer review Not commissioned; externally peer reviewed.