Fatalities in the construction industry: findings from a revision of the BLS Occupational Injury and Illness Classification System

Construction, one of the largest industries in the United States, is also one of the most dangerous. Despite a decline in overall construction injuries thanks to continual prevention and intervention efforts, workers in the industry are still at high risk.¹ In 2011, the construction industry experienced 781 fatal injuries, more than any other industry in the United States.² Accordingly, tracking construction safety and health performance is an important long-term task that will continue to provide updated and accurate information aimed at preventing injuries and illnesses in the industry.

Because of the nature of the work, occupational hazards and exposures in construction are quite different from those in other industries. For example, injuries from falls claim more than one-third of fatalities in construction, accounting for about 40 percent of all work-related fatal falls in the United States.³ As a result, detailed information on falls to a lower level is critical for preventing injuries in construction, but is less relevant to industries which are rarely exposed to hazards that may lead to such falls. Hazards and exposures also vary within construction occupations. For instance, roofers and ironworkers both have a high risk of fatal falls; however, the types of fall-related hazards that they experience on their respective jobsites may be different.⁴ As Joyce Northwood, Eric Sygnatur, and Janice Windau stated, “occupational injuries and illnesses require a context to be best understood.”⁵ Pinpointing specific causes and circumstances that characterize workplace injuries is essential for developing strategies to protect workers from injuries.

Notwithstanding the importance of information underlying falls and other severe workplace injuries, previous data available were insufficient for detailed analysis. For example, despite using data from the Bureau of Labor Statistics (BLS, the Bureau)—the major source for data on occupational injuries and illnesses in the United States—researchers were unable to obtain details regarding fatalities that were due to falls from roofs.⁶ Thus, they failed to identify risk factors for such fatalities and were thereby prevented from formulating strategies that could mitigate those factors. Similarly, questions such as “how many workers died from falls from roofs under 15 feet high?” and “how many pedestrians died from being struck by a vehicle backing up in a roadway or nonroadway area?” were repeatedly asked but remained unanswered for years.

The restructuring of the Occupational Injury and Illness Classification System (OIICS version 2.01) in 2012 has provided a research tool for those who have been longing for detailed information on occupational injuries. OIICS version 2.01 is the first such restructuring since the OIICS was initially released in 1992.⁷ To enhance understanding of the revised coding system, this article uses 2010 and 2011 data from the Census of Fatal Occupational Injuries (CFOI) to analyze fatal injuries in construction. The 2010 data are coded in accordance with the old OIICS, and the 2011 data are coded in conformity with the new, revised OIICS.

Method

The CFOI is conducted by the Bureau through a federal–state cooperative program that has been implemented in all 50 states and the District of Columbia since 1992.⁸ The 2011 CFOI data were coded on the basis of OIICS version 2.01, whereas 2010 CFOI data were classified by the 2007 version of OIICS. Two sections of the coding systems—“Event and Exposure” and “Source”—were selected for analysis. The “Event” section covers the manner in which the incident occurred. For example, fires, falls, and vehicular collisions are all “Events.” The “Source” section includes codes that identify “objects, substances, equipment, and other factors that were responsible for the injury or illness or that precipitated the event.”⁹ For example, chemicals, machinery, tools, and vehicles are all “Sources.”¹⁰ Detailed analysis focused on two major events in construction: falls and transportation incidents. Cross-tabulations were produced for construction fall fatalities by “Primary Source.” The number of fatalities in construction includes all deaths that were coded as belonging in NAICS 23, regardless of whether the worker was self-employed or employed in the private or public sector. These counts may differ from those presented in certain BLS tabulations, which show separate totals for the public and private sectors. The 2010 data presented in this article were generated with SAS version 9.2 under restricted access to BLS CFOI microdata.

Results

Both the new and the old OIICSs have four component structures; two are used to depict the circumstances of the incident (i.e., “Event or Exposure” and “Source of Injury or Illness/Secondary Source”), and two are used to describe the characteristics of the injury or illness (i.e., “Nature” and “Part of Body Affected”). The analysis presented here focuses on the structures and components used to explain the circumstances of the incident. Although both versions of OIICS are organized hierarchically by four-digit numeric codes, the new version eliminated the “0” subsection from “Event” and “Source.” (See tables 1 and 2.) This seemingly minor change has made data manipulation in spreadsheets and statistical packages more efficient without conflating codes. In addition, OIICS 2.01 includes a new “Event” category—Falls, slips, trips—developed by combining Falls with Slip, trip, and loss of balance—without fall from the previous version. (See table 3.) This revision makes sense from an intervention standpoint because prevention methods may be the same for slips and trips, regardless of whether or not they result in a fall. Also, the “Source” categories Containers and Furniture and fixtures, which were separate categories under “Source” in the 2007 version, are combined into one category in version 2.01. (See table 2.) In addition, the order of the “Event” codes has been changed and clearly represents the coding precedence. For example, in the new system, if a worker falls because of a coworker’s push, the “Event” is coded as Intentional injury by person (code 11) rather than Falls, slips, trips (code 4). (See table 1.) The coding would have been similar in the old version; however, the codes were not rank ordered, making their precedence difficult to follow.

With detailed codes for fall injuries in the revised version, the two coding sequences differ substantially in the “Event” section as follows (see table 3):

Version 2.01, “Event or Exposure” component:

Falls, slips, trips (4) → Falls to lower level (43) → Fall through surface or existing opening (432) → 26 to 30 feet (4326).

Version 2007, “Event or Exposure” component:

Falls (1) → Fall to lower level (11) → Fall from roof (115) → Through existing roof          opening (1151).

Table 3 presents fatal falls in construction by detailed “Event” category. The table shows that the category with the greatest number of construction falls was Falls to lower level (2011: n = 260; 2010: n = 256). The third digit in version 2.01 introduces the three major subcategories of Falls to lower level: Fall through surface or existing opening, Fall from collapsing structure or equipment, and Other fall to lower level, into which 72 percent of fatal falls in 2011 were categorized. The fourth digit in version 2.01 provides details about the distance of the fall: Less than 6 feet, 6 to 10 feet, 11 to 15 feet, and so on. About 30 percent of fatal falls were from a height of 15 feet or less. The 2007 version classifies Fall to lower level, such as falling from a ladder, from a roof, and from scaffolding or staging, among others, at the third digit. In 2010, 34 percent of fatal falls in construction were categorized as Falls from roof. Most of the third-digit categories were moved from “Event” to “Source” in version 2.01. According to the new classification rules, for most types of “Event,” a “Source” that “directly produced or inflicted the injury or illness” has been replaced by what is actually responsible for the injury or illness.¹¹ For example, if a person fell from a platform and hit the ground, the old coding system would have listed the “Source” as Ground; in contrast, the new coding system lists the “Source” as Platform. Another data element, the “Secondary Source,” is now available to indicate contributing factors, such as ice or other objects that contributed to a fall. Previously, “Secondary Source” indicated what the worker fell from.

The increase in granularity in OIICS version 2.01 also becomes evident in a comparison of pedestrian vehicular incidents, which are classified under Transportation incidents (2011: n = 221; 2010: n = 209). (See table 4.) An examination of table 4 shows that categories in version 2.01 were expanded to four-digit classification from three-digit classification in the 2007 version. The fourth digit for the 2011 data is more clearly defined and provides more detail than the earlier classification. For example, in the new version, the direction of the vehicle’s motion, such as forward-moving or backing up, is captured. In addition, a separate category for incidents occurring in work zones was added; in 2011, 28 construction workers were killed by forward-moving vehicles in work zones and 14 were killed by vehicles backing up in work zones. In the 2007 version, pedestrian incidents are classified only by the following locations: roadway, side of road, or parking lot/nonroadway area.

Additional information on non-transport-related incidents involving vehicles is available from the category Contact with objects and equipment in version 2.01. Separate categories are available for being struck by a rolling vehicle; a swinging or falling part of a vehicle, such as a boom; a vehicle tipping over; or an object falling from a vehicle.

Table 5 uses fatal falls in construction according to the old and new codes to illustrate the noteworthy changes in “Source.” In OIICS version 2.01, “Source” is categorized by what is responsible for the injury instead of what directly produced the injury, as in OIICS version 2007. By the new codes, Roofs (34 percent) and Ladders (24 percent) were the predominant “Primary Source” of fatal Falls, slips, and trips in construction in 2011. By the old codes, Ground and Floors together were the “Source” of 68 percent of fatal falls in construction in 2010. Similarly, other sources, such as Machinery and Vehicles, were not often used to identify construction falls in the old version, but provide important information in the new version.

Discussion

The coding in OIICS version 2.01 provides more detailed information than that in version 2007, thus enhancing the system’s usefulness in injury prevention. With respect to common fatalities in the construction industry, additional information is available on falls and on pedestrians struck by vehicles, which together accounted for 44 percent of construction fatalities in 2011. Information on the height of falls, along with information on what the worker fell from, can be used in designing fall prevention equipment and developing industry safety standards and best practices. Information on the prevalence of pedestrian fatalities resulting from backward- vs. forward-moving vehicles and in work zones can be used in developing intervention strategies to limit interactions between workers and these vehicles. Finally, the removal of codes with leading zeros makes data manipulation (e.g., data sorting) easier and eliminates confusion between codes, such as might exist between 0319 and 319.

The new coding system does have several disadvantages. One of them is that implementing it increases the difficulty in analyzing data across years. There are nearly 20 years of BLS injury, illness, and fatality data coded according to the 2007 version of OIICS. With the new categories and rules of selection, not all data prior to 2011 are comparable to data from 2011 and forward. Another disadvantage of the new system becomes evident when one is looking at construction falls. The “Event” category Fall to lower level in version 2007 specified primarily what the worker fell from. (See table 3.) Although this information is still available, more than 70 percent of fatalities due to falls in 2011 are categorized in version 2.01 as Other fall to lower level, a category that is too general to be useful. As a result, detailed information on injuries due to falls must be elicited by cross-tabulation of the “Event” and “Source” sections in version 2.01. Moreover, information on the distance the worker fell was not reportable for about 10 percent of the fatal falls to a lower level in construction in 2011. Most of these deaths involved self-employed workers or occurred several months after the initial injury and, therefore, were less likely to have a detailed investigation report describing the incident. This disadvantage affects nonfatal injuries in construction even more: the distance the worker fell was missing for 51 percent of the nonfatal falls to a lower level in private sector construction in 2011.¹² Researchers must use the data with caution in cases where half of the values are missing.

The study presented in this article is based on the first release of data coded according to the newly released OIICS version 2.01. Future studies could include analyzing other “Events” and associated “Sources,” such as Contact with objects and equipment, which accounted for 132 fatal injuries in the construction industry in 2011. Studies could also evaluate detailed data for nonfatal injuries in construction. Once data for several years are classified according to the new OIICS, time-series studies will be possible. Comparing the data before and after the new codes over a longer period than that examined here could give a better idea of the impact of the new codes on safety and health surveillance, as well as be helpful in tracking progress in safety and health in the construction industry.