Abstract
Developing nations bear a substantial portion of the global burden of injury. Public health surveillance models in developing countries should recognize injury risks for all levels of society and all causes and should incorporate various groups of workers and industries, including subsistence agriculture. However, many developing nations do not have an injury registration system; current data collection methods result in gross national undercounts of injuries, failing to distinguish injuries that occur during work. In 2006, we established an active surveillance system in Vietnam’s Xuan Tien commune and investigated potential methods for surveillance of work-related injuries. On the basis of our findings, we recommend a national model for work-related injury surveillance in Vietnam that builds on the existing health surveillance system.
Given the International Labor Organization’s estimate that work-related injuries kill 335 000 people each year (a worldwide rate of 14.0 deaths per 100 000 workers), with developing nations having the highest injury fatality rates,1–3 the lack of detailed health statistics on work-related injuries from developing countries is striking. In Vietnam injuries have grown to be a leading cause of mortality, and in 1996 the Vietnam Ministry of Health (MOH) established a national policy that recognized injuries as a public health problem and resolved to implement community programs centered on localized injury prevention.4 The MOH noted the difficulty in obtaining a comprehensive picture of injury determinants, including workplace injuries, from official statistics. Without accurate data to differentiate injury causes (e.g., traffic vs workplace), the burden of injuries on Vietnam’s economy and their influence on the long-term health and well-being of the country’s residents are largely hidden.
We established an active injury surveillance system in the Xuan Tien commune of Vietnam in 2006 that allowed us to assess a number of potential improvements in surveillance of work injuries. We collaborated with and received the support of numerous levels of health care and government in Vietnam throughout the project (commune, district, province, ministry). The success of these relationships was dependent on the continuous efforts of the research staff at Vietnam’s National Institute for Occupational and Environmental Health. The results of the surveillance project showed that overall injury incidence rates were well in excess of those identified in any prior study in Vietnam, and a large majority of injuries (80%) were judged to be work related. More detailed information on the project is available elsewhere.5–9
Here we apply our findings from the Xuan Tien study to make recommendations on how to improve the means by which existing national health reporting systems in Vietnam track work-related injuries. We use the active surveillance data we collected to examine the likely improvements in data quality that can be realized with incremental changes in the existing health surveillance system. Specifically, we propose adding some of the injury reporting elements we developed in Xuan Tien to the national system and enhancing standard hospital reports to allow more comprehensive collection of data on work-related injuries and their determinants. We describe possible improvements in the sensitivity of data collection on work-related injuries, including collection of information on industries, occupations, and populations at risk; types and causes of injuries; and measures of severity and burden.
WORK-RELATED INJURY SURVEILLANCE MODELS
Enhancement of existing reporting systems in Vietnam to allow identification of work-related injuries is essential to capture the full burden of these injuries and to assist public health authorities in prioritizing risks. Proposals for enhanced surveillance could include either passive or active surveillance approaches. Naturally, sensitivity and accuracy are primary concerns, but cost, feasibility, and the burden on existing institutions and personnel also must be considered.
Although passive surveillance approaches, in which existing records collected administratively (e.g., patient billing records) are used to identify types of injuries and whether they are work related, have the advantage of no added burden, they suffer from the limitations of data generally collected for unrelated purposes (e.g., reduced sensitivity and accuracy). Active surveillance, which involves seeking out additional reports of work-related injuries (e.g., timely collection of reports and administration of additional case questionnaires), would provide more comprehensive and complete information but at a substantial additional cost.
SURVEILLANCE SYSTEM OBJECTIVES
Ultimately, according to the objectives outlined by Souza et al.,10 any national work-related injury surveillance system should be able to provide estimates of the overall magnitude, risk, and severity of work-related injuries at the national and community levels (objective 1); characterize the most common types of work-related injuries (e.g., their nature and cause) and rank them according to burden (objective 2); characterize affected populations (e.g., according to race, gender, age, and socioeconomic position; objective 3); identify geographic areas, industries and occupations, and specific workplaces and communities in which interventions are most needed and rank them according to burden (objective 4); differentiate between national and local priorities (objective 5); and generate support for prevention activities and evaluate the effectiveness of interventions (objective 6).
To understand the current and potential capacity in Vietnam for a work-related injury surveillance system that would accommodate these objectives, we describe the country’s existing health statistics reporting system, including reporting of nonfatal injuries. We then present 3 surveillance models and evaluate them hierarchically, with each model adding components to the preceding model with the goal of improving surveillance of work-related injuries (Table 1). Our approach uses community health treatment sites as the primary source of reporting. We chose not to rely on employer reporting; although larger state-owned and private enterprises have significant occupational medical care resources, including on-site nurses, these companies are few in number relative to the country’s many small- and medium-sized enterprises that rely on the community health care system for occupational as well as all other injury and disease issues.
TABLE 1—
Descriptions of the 3 Surveillance Models Proposed for Improving Surveillance of Work-Related Injuries in Vietnam
| Model | Data Source(s) | Variables Collected/Published in Health Statistics | Case Reportinga | Follow-Up of Injuries | Annual Health Statistics Reports of Work-Related Injuries |
| Enhanced passive surveillance | |||||
| Model 1a | Hospital | Age, gender, address, admission diagnosis, occupation, length of stay | Case variables reported by hospital | None | |
| Model 1b | CHSb | Injury log: occupation, place, reasonb | CHS reportsb | None | Aggregate counts of work-related injuries by occupation |
| Model 2 | CHSb | Discharge diagnosis, activity, location, cause, industry/occupation category, work address (commune)b | Formal quality control of information at district levelb | None | Aggregate counts of work-related injuries distributed by all variables collectedb |
| Active surveillance: model 3 | Active hospital and CHS reporting in districts found from passive surveillance to have high counts of work-related injuriesb | Same as model 2 | Active biweekly collection of hospital and CHS reporting forms in high-risk districtsb | Self-reported work-relatedness, industry, occupation, injury narrative, hours worked, severity, medical cost and burden | Aggregate counts of self-reported work-relatedness, injury types, and outcomes and timely dissemination of findings to the communityb |
Note. CHS = commune health station.
To the district health center, Ministry of Health, and provincial medical center.
In addition to the components listed for the prior model.
REPORTING OF HEALTH STATISTICS IN VIETNAM
Currently, government-run health care facilities throughout Vietnam are required to provide health statistics to district and provincial authorities. The first level of available health care is care from a health volunteer stationed in each hamlet who provides information on vaccinations and other public health programs and offers basic medical treatment. The commune health station (CHS), a permanent facility, provides basic health and emergency care to village residents. Each group of 10 to 20 communes is served by at least 1 district hospital for treatment of more serious cases, and each province has 1 or more provincial hospitals; additional central or specialty referral hospitals are located in major cities.
Local-level administrative and case load information from the CHS is reported through the district health center to the provincial health center and to each subsequent local government authority (commune, district, and province), creating a parallel networked structure of health care and governmental agencies. District and provincial health centers include administrative units dedicated to the design and administration of prevention programs (e.g., programs focusing on AIDS or malaria) and the collection of health survey data and medical statistics. In 2010, there were 45 central-level hospitals, 326 provincial hospitals, 615 district hospitals, and 10 926 CHSs in Vietnam.11
The MOH requires that each of these facilities record and maintain logs of all treatment activity. Admissions logs contain information such as the patient’s name, address, gender, occupation, diagnosis, insurance status, and treatment disposition (e.g., whether the patient was sent home or whether he or she died). Notable with respect to surveillance is the extensive coverage, with 99% of the communes having effective and operational health stations.12 At present, published statistics available for work-related injury surveillance of nonfatal injuries include only cases reported from hospital admissions logs.
Models 1a and 1b: Baseline Passive Surveillance
Model 1a incorporates the administrative data collection procedures currently used in Vietnam. Cases include only those in which patients are treated and admitted to one of the country’s hospitals. In keeping with this model, health staff would be trained to always fill in the diagnosis and occupation data fields, which are often left incomplete in current logs.
Model 1b extends model 1a by integrating published counts from the CHS admissions logs; at present, these data are collected but not published. This model also requires that data on additional variables from the injury patient log (currently collected only at the CHS), including age, gender, address, location at which the injury occurred, occupation, cause of the injury (e.g., traffic accident, occupational accident), and treatment results (recovery, transfer to another hospital, death), be collected at the national level from both hospitals and CHSs. Another requirement is that summary statistics be published on these variables.
Model 2: Enhanced Passive Surveillance
Model 1b and model 2 involve similar data collection strategies (Figure 1). However, in model 2 the injury log variables collected are modified to improve their sensitivity and accuracy in identifying work-related injuries and detailed patient characteristics (Table 1).
FIGURE 1—
Passive and active models for surveillance of work-related injuries.
Note. CHS = commune health station; DHC = district health center; MOH = Ministry of Health; PHC = provincial health center.
The newly defined injury log variables include International Classification of Diseases, 10th Revision (ICD-10) admission diagnosis group codes13; type of activity (engaged in a sport; engaged in leisure; working for pay; working for no pay; going to or coming home from work; resting, sleeping, or eating; doing personal tasks at home, such as taking care of children; school activities; other); location at which the injury occurred (home, residential institution, school, sports facility, street or highway, work, farm, body of water such as a pond or canal); cause of the injury (fall, transport-related accident); industry or occupation at the time of the injury (according to International Labor Organization occupational codes14); ICD-10 discharge diagnosis code, to allow for identification of diagnoses determined after admission (e.g., respiratory disease, back injury); and work address.
If surveillance under model 2 is to be successful, improved compliance with procedures related to collecting and accounting for information will be required. These improvements can be achieved only through increased training of admissions nurses that emphasizes rigorous completion of all log items, directed questions posed before each selection (e.g., “When hurt, were you...working for pay?”), and selection of multiple items from each list when applicable. For example, it is important to know whether an injury occurred at home and the home was the workplace. Data on only some of the required variables may be collectable at admission; the remainder (including injury log details) may need to be collected after treatment but before discharge.
Culturally appropriate translations of ICD-10 and other terminology are necessary (e.g., what is considered “work for pay” or “work for no pay”). The definitions of the required variables must reflect the vocabulary and work experiences of Vietnamese workers and should not be simple translations of definitions used in Western models; development of a detailed variable dictionary is critical.
Model 3: Active Surveillance
Model 3 involves an active surveillance strategy in which timely injury reports at the commune level are gathered from each hospital and CHS, along with follow-up interview data on each case reported. This model includes collection of the same variables as model 2 (on separate injury reporting forms) but offers improvements in data collection and dissemination, including a short lag time between diagnosis and reporting (e.g., biweekly reporting), outreach for improved case reporting, timely feedback to local communities and enterprises, provision of incentives for accurate reporting, and identification of intervention needs and opportunities (Figure 1).
The follow-up interview supplements the data collected in model 2 by gathering information in a narrative form that allows more precise classifications of type of injury, activity, location at which the injury occurred, industry or occupation, and cause of injury. Additional follow-up data collected include self-reported lost work time, family burden (e.g., medical costs), and whether the injury was work related.
Although nationwide application of model 3 is not anticipated, we propose that the model be implemented in certain high-risk districts (identified during passive surveillance) to enhance local prevention and intervention efforts. A representative sample of communes could be established to provide national estimates.
MODEL BENEFITS AND LIMITATIONS
Using published national statistics, local statistics obtained from personal communications with local health administrators at district health centers and CHSs, and our active surveillance data on work-related injuries, we identify some of the benefits and limitations of each model and examine how well the models can satisfy the first 4 objectives of work-related injury surveillance (as outlined earlier).
Sensitivity of Data Collected (Objectives 1 and 2)
Accurate identification of all injury cases is a prerequisite for an accounting of work-related injuries. The health statistics on nonfatal injuries published each year by the MOH (model 1) include aggregations of hospital admissions case data, presented not as counts but as proportionate distributions of reportable conditions. According to these statistics, the second leading cause of death in 2010, as reported by hospitals, was injury and poisoning and other external causes (22%), behind only infectious and parasitic diseases (30%).11
Matsuda15 found that reports from hospitals alone result in gross underestimates of injuries. We found evidence that the current procedures in Vietnam (model 1) lead to underestimates at the local level. In 2004 (the latest year of available data prior to initiation of our study), 1200 injuries, including 105 identified as work related, were reported by the Xuan Truoung district hospital to the district health center, an average of 60 injuries (and 5 work-related injuries) from each of the 20 communes in the district (Xuan Truong District Health Center, unpublished data, 2004). By comparison, during our 2006 study (mentioned earlier),8 intensive active surveillance by the district hospital led to the reporting of 226 injuries from the single commune of Xuan Tien (approximately 4 times higher than the official count in 2004), with 120 of these injuries identified as work related (approximately 24 times higher than the official count).
Another important requirement of surveillance is a case definition that is reasonably sensitive and not overly specific. We found that the English word “injury” is understood by Vietnamese speakers in a much more narrow sense than by English speakers, as including only extreme injuries. When the focus is on more serious injuries, this is appropriate. However, because collection of data on only the most serious injuries will reduce aggregated data, this more narrow definition inhibits the selection of appropriate interventions and fails to measure the total burden of work injuries.
To avoid this situation, we worked with translators and health experts at the National Institute for Occupational and Environmental Health and determined that, of the 3 Vietnamese words that might be understood as injury, the most appropriate would be the Vietnamese noun for hurt. We believe that the low injury counts reported by hospitals and in several published cross-sectional surveys may result from the use of a literal translation of injury.16–20 Similar issues related to language and culture are likely to exist both across and within ICD-10 and other classifications.
In addition, use of official statistics relying on injuries reported only by hospitals will lead to underestimations of even more serious injuries. Specifically, in Xuan Tien we demonstrated that the additional cases reported by the CHS were also very serious and differed with respect to demographic, employment, and cause characteristics (Table 2). In summary, we believe that the current structure (model 1a) entails low sensitivity and significant reporting bias in collection of data on injuries.
TABLE 2—
Sensitivity and Representativeness of Data Sources: Results From Active Surveillance of Work-Related Injuries in Xuan Tien Commune, Vietnam, January–December 2006
| All Work-Related Cases Reported |
District Hospital and Commune Health Station |
District Hospital Only |
||||
| Characteristic | No. (%) | Mean No. of Days Losta | No. (%) | Mean No. of Days Losta | No. (%) | Mean No. of Days Losta |
| Age, y | ||||||
| < 15 | 8 (2) | 16 | 2 (1) | 2 (2) | ||
| 15–24 | 104 (21) | 12 | 84 (25) | 9 | 20 (19) | 15 |
| 25–44 | 212 (42) | 10 | 140 (41) | 11 | 40 (38) | 19 |
| 45–64 | 150 (30) | 11 | 92 (27) | 10 | 38 (36) | 16 |
| ≥ 65 | 29 (6) | 13 | 23 (7) | 16 | 5 (5) | 18 |
| Missing | 1 (0) | |||||
| Gender | ||||||
| Male | 309 (61) | 11 | 221 (65) | 11 | 58 (55) | 18 |
| Female | 195 (39) | 11 | 120 (35) | 11 | 47 (45) | 15 |
| Nature of injury | ||||||
| Open wound | 246 (48) | 7 | 167 (49) | 7 | 15 (14) | 12 |
| Sprain or strain | 71 (14) | 7 | 49 (14) | 6 | 31 (30) | 7 |
| Superficial injury | 57 (11) | 8 | 38 (11) | 9 | 17 (16) | 10 |
| Other unspecified | 33 (6) | 8 | 25 (7) | 11 | 9 (9) | 11 |
| Fracture | 18 (4) | 37 | 15 (4) | 39 | 13 (12) | 45 |
| Burn | 17 (3) | 7 | 5 (1) | 10 | 1 (1) | 15 |
| Contusion | 15 (3) | 10 | 11 (3) | 10 | 5 (5) | 7 |
| Pain specified | 11 (2) | 2 | 7 (2) | 5 | 5 (5) | 5 |
| Nerve | 7 (1) | 7 | 7 (2) | 7 | 2 (2) | |
| Foreign body | 8 (2) | 2 | 6 (2) | 2 | 0 (0) | |
| Crushing injury | 6 (1) | 30 | 3 (1) | 19 | 0 (0) | |
| Other specified | 5 (1) | 10 | 4 (1) | 11 | 3 (3) | 14 |
| Amputation | 4 (1) | 12 | 2 (1) | 2 (2) | ||
| External effect | 2 (0) | 2 (1) | 1 (1) | |||
| Dislocation | 3 (1) | 7 | 1 (0) | 1 (1) | ||
| Unclassifiable | 1 (0) | |||||
| Industryb | ||||||
| Manufacturing | 270 (54) | 11 | 202 (59) | 9 | 46 (43) | 17 |
| Agriculture | 113 (22) | 10 | 65 (19) | 10 | 24 (23) | 15 |
| Wholesale/retail | 46 (9) | 15 | 33 (10) | 17 | 16 (15) | 22 |
| Construction | 33 (7) | 11 | 21 (6) | 12 | 8 (8) | 18 |
| Other classifiable | 22 (4) | 8 | 12 (4) | 19 | 7 (7) | 12 |
| Unclassifiable | 20 (4) | 8 (2) | 4 (4) | |||
| Cause (ICD-10 category)b | ||||||
| Contact | 155 (31) | 8 | 105 (31) | 8 | 11 (10) | 21 |
| Overexertion: other | 92 (18) | 7 | 66 (19) | 6 | 35 (33) | 6 |
| Machinery | 56 (11) | 16 | 45 (13) | 11 | 6 (6) | 18 |
| Fall | 48 (10) | 18 | 32 (9) | 21 | 19 (18) | 26 |
| Motor vehicle | 20 (4) | 26 | 15 (4) | 29 | 6 (6) | 44 |
| Other transport | 37 (7) | 10 | 26 (8) | 11 | 15 (14) | 12 |
| Cutting/piercing | 28 (6) | 7 | 11 (3) | 9 | 1 (1) | |
| Hot substances | 10 (2) | 8 | 1 (0) | 0 (0) | ||
| Foreign body | 9 (2) | 3 | 7 (2) | 3 | 0 (0) | |
| Fire/flames | 5 (1) | 6 | 2 (1) | 1 (1) | ||
| Other specified | 22 (4) | 18 (5) | 6 (6) | |||
| Unclassifiable | 22 (4) | 13 (4) | 5 (5) | |||
| Total | 504 (100) | 11 | 341 (100) | 11 | 105 (100) | 17 |
Note. ICD-10 = International Classification of Diseases, 10th Revision.
Among those who lost time. No means reported for fewer than 3 cases.
Includes hospital, commune health station, and health volunteer (to simulate private care) reports of work-related injuries. Categories with at least 5 cases overall are included; the remainder are aggregated in the “other specified” category.
Models 1b, 2, and 3 partially address this issue by allowing additional CHS cases to be reported. Also, use of model 2 would improve detection of injured patients through mandatory ICD-10 diagnosis assignments for each admitted patient. Nevertheless, a significant undercount of injury cases is still to be expected with any passive reporting surveillance system; health staff will attempt to collect diagnosis information as part of the larger admissions and billing process when their focus at that point will clearly be on triaging patients for treatment and administrative procedures. However, although identifying as many cases as possible is important to allow comparisons with other conditions, we believe that model 2 can be used to collect a representative case load sufficient to offer an understanding of appropriate national case characteristics.
Periodic training on reporting guidelines for nurses and doctors at both hospitals and CHSs, along with timely recognition and reporting of injuries, will substantially improve case reporting sensitivity as well as data integrity (model 3). Active reporting may also create community concern for and awareness of injury risks similar to the situation with infectious diseases.
Work-Relatedness and Determination of Targets (Objectives 1 and 4)
At present, hospitals in Vietnam have no MOH-mandated reporting procedures to identify work-related injuries (model 1). Baseline injury information from the CHS in Xuan Tien for 2004 (model 1b) was requested and provided (Table 3). A total of 97 injuries (26%) were reported as injuries to workers, 82 (22%) were classified as work accidents, and 79 (21%) occurred at workplaces. However, in our 2006 study in Xuan Tien, we identified more than 300 work-related injuries (80% of all injuries) that were actively reported from the CHS alone. These results suggest that using the MOH procedures would significantly underestimate the true numbers and percentages of work-related injuries occurring annually in Xuan Tien.
TABLE 3—
Informally Reported Injury Cases Before Study Initiation: Xuan Tien Commune, Vietnam, 2004
| Cases (n = 379), No. (%) | |
| Occupation | |
| Civil servants | 6 (2) |
| Farmers | 58 (15) |
| Army personnel | 2 (1) |
| Students | 186 (49) |
| Workers | 97 (26) |
| Free laborers | 15 (4) |
| Other | 15 (4) |
| Location at which injury occurred | |
| On the road | 162 (43) |
| Home | 72 (19) |
| School | 59 (16) |
| Workplace | 79 (21) |
| Public area | 2 (1) |
| Pond | 5 (1) |
| Cause | |
| Traffic accident | 159 (42) |
| Work accident | 82 (22) |
| Drowning | 5 (1) |
| Burning | 18 (5) |
| Suicide | 1 (0) |
| Fighting | 13 (3) |
| Other | 101 (27) |
| Location of treatment | |
| Home | 30 (8) |
| Commune health station | 263 (69) |
| Higher-level facility | 86 (23) |
Our findings suggest that, in the current data collection process, occupational titles lack sufficient specificity and provide insufficient information on work exposures and work-relatedness. If more specific occupational titles were identified, along with evidence that a patient was injured while at work, work-relatedness could be established. In addition, eliminating the use of forced mutually exclusive categories (e.g., the workplace and home are frequently the same location) would allow a more complete log of work injuries.
Therefore, model 2 expands data collection with variables for activity, location at which the injury occurred, industry, and occupation. This should substantially improve identification of work-related injury cases and allow for better estimations of the magnitude of these injuries and associated mechanisms in Vietnam. With passive reporting, however, low completion rates of the new variable fields will remain a concern.
Only active reporting (model 3) will yield results similar to those outlined in Table 2. This data collection approach proved successful with respect to comprehensive assignment of work-related injuries in the Xuan Tien commune.8 Accuracy in the assignment of work-related injuries is expected to improve with collection of separate injury reporting forms and administration of follow-up surveys that include additional direct questions on the circumstances of the injury and whether it is work related.
Populations at Risk and Common Injury Types (Objectives 2 and 3)
Although age, gender, address, and occupation are included as items in the current admission logs, the address and occupation fields are often left blank. Models 2 and 3 specify that home and work addresses be included as separate entries, which will help in the selection of geographical targets. These 2 models also incorporate specific ICD-10 nature of injury codes and cause of injury reports (Table 2).
Severity and Burden of Work-Related Injuries (Objectives 1 and 4)
In both models 2 and 3, ICD-10 diagnoses (e.g., amputation vs laceration) can be used as a severity metric. However, only model 3 is capable of collecting additional data on indicators of burden, such as lost work time and medical costs. These data are critical in gaining an understanding of the social and economic benefits to a community of improving workplace safety and ranking priorities.
SUMMARY OF MODEL ADEQUACY AND FEASIBILITY
Although model 1 is the most cost-efficient and simplest to implement, the sensitivity of this model in detecting injuries is low, and it includes biases. In addition, there is no clear path to identifying work-related injuries and no way to accurately describe populations at risk.
Model 2 requires small structural changes to the current system and additional training of staff. If rigorous data collection protocols and systems are established to ascertain ICD-10 diagnoses and whether an injury is work related, model 2 can potentially identify injury types, demographic characteristics, and geographical target areas, as well as national-level priorities in terms of magnitude of work-related injuries. Although model 2 should involve less misclassification than model 1 and a more comprehensive description of case characteristics, the completeness, accuracy, and validity of the information collected are likely to be compromised with passive reporting; details such as diagnosis and injury log data may not be considered important or available during admission and may be forgotten later. Therefore, we believe that this model has improved but limited value with respect to determining priorities and focusing interventions at the local level, which can have an impact in Vietnam’s many small- and medium-sized enterprises.
Model 3 is capable of satisfying almost all of the objectives of surveillance of work-related injuries. Active surveillance procedures could be successfully implemented in rural communes where the structure of family and work life is similar to that of Xuan Tien and there is a general commitment to improve working conditions. Significant resources would be needed, however, for training, support of ongoing reporting procedures, follow-up survey administration, and data management. Although it is unlikely that nationwide active surveillance is possible at this stage of Vietnam’s economic development, the approach outlined in model 3 could be used to identify a representative sample of communes to report surveillance results as a means of informing national policies.
SUMMARY
Integrating a system for national surveillance of work-related injuries in Vietnam is necessary to establish the true magnitude and causes of these injuries and generate political support for the resources necessary to reduce the associated burden. We believe that Vietnam has the capability, with modifications of the country’s current health statistics reporting system, to be successful in determining the incidence and burden of work-related injury.
We recommend that an extended passive surveillance approach be adopted in Vietnam that would include hospital and CHS reporting (model 2). Reporting logs would be modified to collect discharge diagnosis ICD-10 codes and several specific variables that would establish, in a majority of cases, whether an injury was work related. In addition, rigorous training procedures should be developed and administered to health staff to improve the collection of demographic information and data on admissions. We believe that these changes alone would lead to substantial improvements in terms of satisfying the objectives of work-related injury surveillance.
As health authorities become aware of counts or rates in specific communities contributing unduly to the national injury burden, active surveillance in those communities (model 3) would be a valuable extension to identify certain workplaces, communities, or industries for targeted improvements. Such targeted efforts would also assist the small- and medium-sized enterprises characteristic of industrializing economies that require more than national statistics alone. The majority of enterprises in Vietnam currently employ 10 to 20 workers and are frequently home-based businesses. Given their scarce resources, enterprise owners need detailed local information that is not available in national statistics to prioritize interventions and address hazards in the most effective way.
As in many countries worldwide, determination of the impact of work-related injuries in Vietnam involves both health and labor authorities. Currently, estimates of the scale of occupational injuries are based solely on employers’ reports. The need to collect work-related injury statistics at the national level from health records may not be readily apparent in Vietnam because of the existence of regulations requiring that employers report workplace injuries to the Ministry of Labor Invalids and Social Affairs.20 According to occupational health leaders at the country’s highest levels of government, however, workplace injuries are substantially underestimated.21 The proposed engagement of the MOH in surveillance of work-related injuries would facilitate collaborations with the Ministry of Labor Invalids and Social Affairs on a major issue that contributes to Vietnam’s health and economic burden.
Even if the current surveillance system is enhanced as recommended, completeness of data capture is a concern when only 2 data sources (hospitals and CHSs) are used. Although many state-run health care facilities are available, studies suggest that some injured workers will self-treat or seek out private care because of lack of insurance coverage, lack of resources to pay for treatment, or lack of available care.12,22,23 Thus, injuries will be undercounted if state facilities are considered the sole providers of injury reports.24 These undercounts could be substantial in certain communities; Hang and Byass23 found that, in Vietnam’s Fila Bavi district, 77% of all injuries were self-treated or treated by private care physicians.
In addition, whereas the CHS is the primary provider of treatment in some rural communes (e.g., Xuan Tien), this may not be the case in other communes or cities. Therefore, additional sampling of private care physicians or other innovative approaches to collecting data on informally treated injuries may be necessary in some communities to increase sensitivity and provide the data necessary to identify dangerous processes and specific circumstances or workplace conditions that lead to injuries.25 Our previous investigation indicated that gathering injury reports based on informal first aid treatments at workplaces could identify more work-related injuries as well as practical improvements that could be made by various work sites and industries. Involving enterprise owners in this surveillance was also instrumental in raising awareness of problems in their own workplaces.8
Although interventions designed to mitigate communicable diseases (e.g., quarantines, chlorination of water supplies) are usually established by public authorities and supported by both the government and the general public, surveillance of occupational injuries requires support and action on the part of employers. In developing communities such as Xuan Tien, industrialization is still very local and intertwined with family life (workers and owners are from the same community, often the same family), which leads to a local environment where the decisions of enterprise owners are moderated by the social environment. Our experience suggests that there may be an exceptional opportunity to build community support for local active surveillance of work-related injuries as well as commitments from employers to establish interventions in the communities in which they operate.
Acknowledgments
We thank Vu Viet Hung, the nurses from the Xuan Truong district hospital, and all other commune, district, and provincial party leaders and dedicated health staff who supported our work. We also thank Ngo Quynh Sang, the main physician and coordinator for the commune study, and Tran Hai Yen, who bridged between-culture gaps in understanding. We thank William Shaw and Larry Hettinger for insightful reviews. Finally, we thank Peg Rothwell for editorial input.
Human Participant Protection
The institutional review boards at the Liberty Mutual Research Institute for Safety; the University of Massachusetts, Lowell; and the Vietnam National Institute for Occupational and Environmental Health approved this study. We obtained written informed consent from the head of household or most senior person living in each household taking part in the survey.
References
- 1.Concha-Barrientos M, Nelson DI, Fingerhut M, Driscoll T, Leigh J. The global burden due to occupational injury. Am J Ind Med. 2005;48(6):470–481. doi: 10.1002/ajim.20226. [DOI] [PubMed] [Google Scholar]
- 2.Herbert R, Landrigan PJ. Work-related death: a continuing epidemic. Am J Public Health. 2000;90(4):541–545. doi: 10.2105/ajph.90.4.541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Horan JM, Mallonee S. Injury surveillance. Epidemiol Rev. 2003;25:24–42. doi: 10.1093/epirev/mxg010. [DOI] [PubMed] [Google Scholar]
- 4.Luau HC, Svanström L, Ekman R et al. Development of a national injury prevention/safe community programme in Vietnam. Health Promot Int. 2001;16(1):47–54. doi: 10.1093/heapro/16.1.47. [DOI] [PubMed] [Google Scholar]
- 5.Marucci-Wellman H, Leamon TB, Binh TT, Diep NB, Wegman DH, Kriebel D. A survey of work-related injury in a rapidly industrializing commune in Viet Nam. Int J Occup Environ Health. 2009;15(1):1–8. doi: 10.1179/107735209799449734. [DOI] [PubMed] [Google Scholar]
- 6.Marucci-Wellman H, Leamon TB, Binh TT et al. The work-related burden of injury in a rapidly industrialising commune in Viet Nam. Occup Environ Med. 2010;67(4):244–250. doi: 10.1136/oem.2009.046227. [DOI] [PubMed] [Google Scholar]
- 7.Marucci-Wellman H, Leamon TB, Willetts JL et al. Occupational injuries in a commune in rural Vietnam transitioning from agriculture to new industries. Am J Public Health. 2011;101(5):854–860. doi: 10.2105/AJPH.2010.300019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Marucci-Wellman H, Leamon TB, Binh TT, Diep NB, Wegman DH, Kriebel D. The burden of occupational injury: a 1-year prospective study in Xuan Tien commune, Viet Nam. Am J Ind Med. 2012;55(3):205–216. doi: 10.1002/ajim.21039. [DOI] [PubMed] [Google Scholar]
- 9.Marucci-Wellman H, Willetts JL, Diep NB, Binh TT. The utility of injury narratives for prioritising future prevention activities in small and medium enterprises in rural Vietnam. Inj Prev. 2013;19(2):92–99. doi: 10.1136/injuryprev-2012-040368. [DOI] [PubMed] [Google Scholar]
- 10.Souza K, Davis L, Shire J. Environmental health surveillance. In: Levy BS, Wegman DH, Baron SL, Sokas RK, editors. Occupational and Environmental Health: Recognizing and Preventing Disease and Injury. 6th ed. New York, NY: Oxford University Press; 2011. pp. 55–68. [Google Scholar]
- 11.Health Statistic Yearbook 2010. Hanoi, Vietnam: Ministry of Health; 2011. [Google Scholar]
- 12.Adams SJ. Vietnam’s Health Care System: A Macroeconomic Perspective. Tokyo, Japan: Hitotsubashi University; 2005. [Google Scholar]
- 13.International Statistical Classification of Diseases and Related Health Problems, 10th Revision. Geneva, Switzerland: World Health Organization; 1992. [Google Scholar]
- 14. International Labour Organization. Resolution concerning statistics of occupational injuries (resulting from occupational accidents). Available at: http://www.ilo.org/global/statistics-and-databases/standards-and-guidelines/resolutions-adopted-by-international-conferences-of-labour-statisticians/WCMS_087528/lang–en/index.htm. Accessed April 21, 2013.
- 15.Matsuda S. An analysis of the Vietnamese system of occupational safety and health and setting priorities with the analytical hierarchy process. Occup Environ Med. 1996;53(4):281–286. doi: 10.1136/oem.53.4.281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hang HM, Ekman R, Bach TT, Byass P, Svanstrom L. Community-based assessment of unintentional injuries: a pilot study in rural Vietnam. Scand J Public Health Suppl. 2003;62:38–44. doi: 10.1080/14034950310015095. [DOI] [PubMed] [Google Scholar]
- 17.Linh L, Anh L, Linnen M, Cuong P. Burden of injury in Vietnam: first national survey. Paper presented at: 7th World Conference on Injury Prevention and Safety Promotion, June 6–9, 2004, Hanoi, Vietnam. [Google Scholar]
- 18.Phung DT, Nguyen HT, Mock C, Keifer M. Occupational injuries reported in a population-based injury survey in Vietnam. Int J Occup Environ Health. 2007;14(1):35–44. doi: 10.1179/oeh.2008.14.1.35. [DOI] [PubMed] [Google Scholar]
- 19.Nguyen HT, Pham CV, Keifer M, Mock C. Occupational injuries as reported in the Vietnam Multi-Center Injury Survey. Asia Pac J Public Health. 2008;20(suppl):196–202. [PubMed] [Google Scholar]
- 20.National Programme on Occupational Safety and Occupational Health in Period of 2011–2015. Hanoi, Vietnam: Ministry of Labour Invalids and Social Affairs; 2012. [Google Scholar]
- 21.Luong P. Occupational Accidents in Vietnam—Situation and Solution. Hanoi, Vietnam: Vietnam Occupational Health and Safety Association; 2007. [Google Scholar]
- 22.Giang TL. Social Health Insurance in Vietnam: Current Issues and Policy Recommendations. Bangkok, Thailand: International Labor Organization Subregional Office for East Asia; 2008. [Google Scholar]
- 23.Hang HM, Byass P. Difficulties in getting treatment for injuries in rural Vietnam. Public Health. 2009;123(1):58–65. doi: 10.1016/j.puhe.2008.07.018. [DOI] [PubMed] [Google Scholar]
- 24.Erhart A, Thang ND, Xa NX et al. Accuracy of the health information system on malaria surveillance in Vietnam. Trans R Soc Trop Med Hyg. 2007;101(3):216–225. doi: 10.1016/j.trstmh.2006.07.003. [DOI] [PubMed] [Google Scholar]
- 25.Lipscomb HJ, Glazner JE, Bondy J, Lezotte D, Guarini K. Analysis of text from injury reports improves understanding of construction falls. J Occup Environ Med. 2004;46(11):1166–1173. doi: 10.1097/01.jom.0000141769.48553.1b. [DOI] [PubMed] [Google Scholar]