LMRIS work summary by Gary S. Sorock, Ph.D
I worked at the LMRIS from 1993-2003 where I followed three lines of research. They were narrative text analysis, case crossover studies, and work-related injuries in China. I briefly sketch each research area. Then I end each with a researchable question for future researchers to explore.
When my new boss, Dr. Tom Leamon, implied I would have no direct supervisor at the LMRIS, I wondered if it was because he had no clue what his newly-hired injury epidemiologist could do. I learned sometime later that he did in fact know. He partnered with the Harvard School of Public Health and other epidemiologists. But at the outset, what was I to do as a new employee surrounded by a biomechanical engineer and a few ergonomists? Tom suggested speaking to a LM Loss Prevention Transportation Specialist in Boston. The Safety Specialist asked a useful question.
Narrative Text Analysis
Could we, he asked, “use LM claims data to determine the circumstances of motor vehicle crashes in roadway construction workzones?” This time I had no clue how to proceed. I did do some database research on my previous job and quickly (no, painstakingly slowly) got access to the LM claims database. I downloaded an extract of all recent auto claims with the keywords “construction workzone” from the crash narrative supplied by the claimant when interviewed by telephone. A few other researchers at NIOSH were also analyzing “words as data.” It was the only way to analyze un-coded data from a large database. This line of research led to a publication1 and to more inquiries into machine learning of coding claims data using narrative data2. I wonder if computer software can be designed to add a line of follow-up inquiry about possible causal risk factors of crashes when the claim intake process occurs? This it seems would involve close collaboration between epidemiologists, software developers, and claims intake managers. It might add to Loss Prevention’s expertise in automobile crash prevention.
The blending of two disciplines: human factors and epidemiology led, in my view, to the most substantive contribution of Liberty’s work on injury prevention. Human factors literature on situational awareness and fault-tree analysis provided important theoretical causal risk factors like distraction, equipment malfunction and rushing. Epidemiology provided the operational means to measure the real-world risk of an injury when exposed versus unexposed to these potential risk factors. The case-crossover design was originally used to study transient risk factors for sudden onset heart attacks. We applied this research design to study short-term transient risk factors for acute hand trauma at work. In our telephone-interview based study of acute hand-injured workers (n=1,166) (mostly lacerations or crush injuries), doing an unusual task, using malfunctioning equipment or tools, using a different method than usual to do a task and distraction and rushing were all significant risk factors for acute hand trauma across occupational category, work experience and safety training3. The same methodology could be used in other acute-injury outcome events that are frequent enough and that have a known time of onset. However, cumulative wrist injury, insidious onset back pain or infrequent events like mid-sea collisions of military and commercial ships may not be amenable to study using the case-crossover design. In this line of research, the most interesting question came from a “blind” reviewer of our early results manuscript: The reviewer used the term “tautology” to describe our research. A tautology is a needless repetition of an idea, e.g., “she was a dark-haired brunette.” The question assumes that acute injury is an unusual event therefore studying unusual precursors is a tautology and presumably worthless. I responded by writing how epidemiologists since the early days of John Snow looked for whatever means he could find to decrease the risk of cholera in London in 1885. We were engaged, I wrote, in the same type of endeavor.
Identifying unusual exposures when the risk of hand injury were increased or decreased led us to further our efforts to suggest hand injury prevention strategies. The notion of unusual exposures prior to “accidents” often leads me even today to search for such exposures in apparent singular event “accidents.” Consider the Challenger explosion, the 9/11 attacks or the recent Amtrak crash outside of Philadelphia. The first event involved ‘O’ ring performance in colder-than-usual weather, the second by flight training of multiple men at the same time from countries in the middle east and the third by a rock hitting the window of a train engineer’s cab. All unusual exposures and all could have increased suspicion of an impending untoward event. Interestingly, I spoke with a NASA test pilot years ago about this point. He said whenever anything occurs that is “unusual” with the aircraft or in the cockpit, his awareness of the increased risk of further unusual exposures or potential “accidents” is heightened. Even this situational awareness may not be enough though when a sudden flock of geese appear in front of a commercial jet aircraft damaging the jet engines. Yet flocks of birds can appear and be tracked on radar. Future studies are needed to determine how best to intervene in a situation when an unusual exposure occurs. In other words: How can situational awareness of transient risk factors be increased in order to prevent rare events? In other words, read the roadway sign: “Don’t text and drive.”
Work-Related Injuries in China
I was indeed fortunate that Liberty Mutual wanted to share our knowledge with the world. This led to collaborations with researchers from The Netherlands, India, Australia and to China. In the case of China, we learned that international research takes much, much longer to conduct partially because of the cultural barriers. Building trust does not happen with one four-day visit to Shanghai. Learning some Mandarin, and building post-doctoral programs between Fudan University in Shanghai and Harvard University in Boston was another crucial building block for this success. Over many years for example, we were able to replicate U.S. findings regarding fall-related fatalities in the construction industry in Shanghai and the value of perimeter guarding4. The case-crossover study design of occupational injury was introduced into China and continues to be used there and elsewhere in Europe5,6. My question is: Can a world-wide collaboration produce a software application on a cell phone device that can capture case-crossover data in multiple languages? This might begin with occupational hand trauma and then expand to motor-vehicle crashes. The results of these studies and safety warnings might be available in the same application on a cell phone from multiple countries.
1.Sorock G, Ranney T, Lehto M. Motor-vehicle crashes in roadway construction workzones: An analysis using narrative text from insurance claims. Accident Analysis and Prevention. 1996: 28: 131-138.
2.Lehto M, Sorock G. Machine learning of motor vehicle accident categories from narrative data. Methods of Information in Medicine 1996: 35: 309-316.
3. Sorock GS, Lombardi DA, Hauser R, Eisen EA, Herrick B, Mittleman MA. A case-crossover study of transient risk factors for occupational acute hand injury. Occupational Environmental Medicine 2004; 61: 305-361
- Xia J, Sorock GS., Zhu, J., Courtney T., Fu H., Liang Y., Christiani D. Fatal occupational injuries in the construction industry of a new development area in East China, 1991-1997. American Industrial Hygiene Association J, 2000; 61: 733-737.
- Jin K, Lombardi D, Courtney T, Sorock G, Perry M, Chen H, Wang X, Liang Y. The challenge of cross-cultural collaborative research: lessons learned from a pilot case-crossover study of severe occupational hand trauma in the People’s Republic of China. Injury Prevention 2007; 13: 133-6).
- Lombardi D. Advances in occupational traumatic injury research. The Scandanavian Journal of Work Environment and Health 2017 43(3): 191-195.
This summary is dedicated to the memory of Bill Patterson, M.D. who was the premier recruiter of hand-injured patients in our U.S. based study. Bill was the Medical Director of the Occupational Health and Rehabilitation Inc. clinic in Wilmington, MA, in the above mentioned hand-injury study operated out of the LMRIS in Hopkinton, MA, USA.