| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH |
|
|
|
|||||||
|
|||||||||
TRACK to:
|
|
Electronic letters published:
![[Read Comment]](/icons/misc/sectionDOWN.gif){kind=icon}
Further subgroup analyses
An epidemic of injury and and epidemic of anger
Re: Anger and injury
Anger and injury
|
|
|||||||||||||||
|
Daniel C. Vinson, Columbia, MO Family and Community Medicine, University of Missouri-Columbia
Send response to journal:
|
Doing many subgroup analyses increases the risk of Type 1 errors, but Dr. Zolotor’s question is intriguing. We examined the anger-injury association by injury severity. Among the 1678 patients with minor injuries (defined as an Injury Severity Scale[1,2] score of 1), the odds ratio in case-crossover analysis for each point on the 0 to 12-point scale created by summing the scores for irritable, angry, and hostile was 1.13 (95% CI 1.09 to 1.18). Among those with more serious injuries (N = 827), the OR was 1.19 (1.10 to 1.28). Examining just hostile, the odds ratio per point on the 0 to 4 scale was 1.44 (1.25 to 1.66) with minor injury, 1.64 (1.27 to 2.13) with more serious injury. Maybe there’s an association, but even with this large sample, it’s not statistically significant. Dr. Zolotor asked about sociodemographic variables. Ethnicity, education, and insurance status were associated with injury risk and with anger, but controlling for them in case-control analyses changes the odds ratio very little. Dan Vinson, M.D. REFERENCES 1. Vinson DC, Maclure M, Reidinger C, Smith GS. A population-based case-crossover and case-control study of alcohol and the risk of injury. J Stud Alcohol. 2003;64(3):358-366. 2. Association for the Advancement of Automotive Medicine. Abbreviated injury scale: 1990 revision. Des Plaines, Il.: Association for the Advancement of Automotive Medicine; 1990. Competing interests: None declared |
|||||||||||||||
|
|
|||||||||||||||
|
Adam J. Zolotor, Chapel Hill, NC Instructor, Departmentr of Family Medicine, Univerisity of North Carolina at Chapel Hill
Send response to journal:
|
Vinson and Arelli’s paper “State anger and the risk of injury: a case -control and case crossover study,” makes an important contribution to understanding the relationship between anger and injury. The combination of case-control and case-crossover designs is particularly convincing. In this study, case-control design is limited by the use of few demographic control variables. Risk of injury is positively associated with low economic status and minority race or ethnicity(1, 2). In addition, these are associated with emergency department utilization(3). Also, they are all negatively associated with having a land-lined telephone (4). If any of these characteristics are also associated with state anger, then the association of state anger with injury will be overstated. The use of case-crossover design matches the respondent’s self-reported state anger just before injury with self-reported anger 24 hours prior to injury. This will effectively eliminate any bias due to race, ethnicity, or economic status. The consistent result between design types is a true strength of this study. Also, the plausibility of this relationship and the dose-response indicated by table 3 provide fairly compelling evidence that this relationship is not spurious. As the authors indicate, recall bias may obscure the relationship between state anger and injury. I appreciate their post-publication analysis of the association between other negative emotions and injury, which are mostly positive associations. As Dr. Sonis points out, this could be due to recall bias or a global association of negative emotions with injury. Given that ‘state’ is a moment to moment phenomenon, it is hard to imagine a prospective cohort study testing this association. If possible, it would be informative to look at injury severity in the context of this study. If more severe injuries are associated with more state anger, it would add another dimension to the dose-response relationship. Also, more information about injury severity may bolster the public health implications of the current study. I find the rates of hostility, anger, and irritability in the community-based controls alarming. Given that these state emotions are at epidemic proportion and that injury causes over 150,000 deaths and 30 to 40 million emergency department visits each year(5), Vinson and Arelli may be pointing to an important social phenomenon that may have far reaching health and societal costs. 1. National Center for Injury Prevention and Control. Injury Fact Book 2001-2002. Atlanta, GA: Cenders for Disease Control; 2002. 2. Aber JL, Bennett NG, Conley DC, Li J. The effects of poverty on child health and development. Annu Rev Public Health 1997;18:463-83. 3. Gill JM, Fagan HB, Townsend B, Mainous AG, 3rd. Impact of providing a medical home to the uninsured: evaluation of a statewide program. J Health Care Poor Underserved 2005;16(3):515-35. 4. Salant D, Dillman DA. Choosing a survey method. In: How to Conduct Your Own Survey. New York, NY: John Wiley & sons, Inc.; 1994. 5. National Center for Injury Prevention and Control. CDC Acute Injury Care Research Agenda: Guiding Research for the Future. Atlanta, GA: Centers for Disease Control; 2002 Competing interests: None declared |
|||||||||||||||
|
|
|||||||||||||||
|
Daniel C. Vinson, MD, MSPH, Columbia, MO, USA Professor, University of Missouri-Columbia, Vineesha Arelli, BS
Send response to journal:
|
We appreciate Dr. Sonis’s thoughtful critique and questions. As Dr. Sonis points out, anger in our study might be associated with traffic injuries, and our inability to demonstrate it may simply be because of the small sample size. We interviewed 600 patients presenting with traffic injuries, not a small sample at all; but only 14 reported they were “extremely” “angry” just before the injury event. Because 14 is a relatively small number, the confidence intervals in both case-crossover and case-control analyses are wide and include 1. We did not say that anger was not associated with non-intentional injuries. Indeed it was. It’s just that the association with intentional injury was far stronger than for other injury mechanisms. Prompted by his question, we’ve done further analyses. We collected data on “sad,” reported for both injury time and matched control time using the same 5-point Likert scale as “angry” and the other emotions we examined. Each point on that 5-point scale had an odds ratio of 1.13 in case-crossover analysis, and 0.95 in case-control analysis. Words related to fear (“afraid,” “scared,” “frightened”) were associated with injury risk in both case-crossover and case-control analyses, but, as Dr. Sonis points out, those associations could be because of recall bias or because these negative emotions are truly associated with injury. The latter explanation is plausible, though the connection might not be causal (a risky situation, for example, could cause both fear and increased injury risk). The table here shows odds ratios and 95% confidence intervals for these four negative emotions.
Clearly, further research more focused than our study on the specific relationships between anger and injury is needed. Competing interests: None declared |
|||||||||||||||
|
|
|||||||||||||||
|
Jeffrey H Sonis, Chapel Hill, NC, USA Assistant Professor, Department of Social Medicine, University of North Carolina at Chapel Hill
Send response to journal:
|
Vincent and Arelli report that “state anger”, i.e. transient angry emotion, increases the odds of any injury and intentional injury. These are interesting and, if true, important findings. Their article has important strengths. First, the use of a case- crossover design, in which each person served as his/her own control, eliminates the possibility of confounding by stable personality and demographic characteristics. Second, the pattern of associations between anger and injury was similar in the case-crossover and the case-control studies. Third, there was a strong dose-response relation between anger and any injury. Fourth, the findings are plausible, because arguments are the most common precipitating cause of violence.1 However, I have several concerns. The most serious problem is the potential for recall bias in the assessment of anger. If persons who were injured attributed their injury to negative emotions, and those beliefs influenced their retrospective assessment of their emotions at the time of the injury, the odds ratios for the anger / injury association would be falsely elevated. The authors could help us assess the “recall bias” hypothesis by reporting the associations between the other negative emotions (e.g., “scared”) that were assessed in the study and injury. If other emotions show associations with injury that are similar to anger, then either recall bias produced the findings, or negative emotions globally are associated with injury. I also disagree with their conclusion that the reported data showed that anger was associated with intentional but not non-intentional injuries. They cite as evidence against an association between extreme anger and traffic injuries an odds ratio of 4.2 (95% CI 0.9 - 20) in the case-crossover study and an odds ratio of 4.6 (95% CI, 0.8 – 28) in the case-control study. However, since the odds ratios are roughly comparable to those for extreme anger in which “any injury” was the outcome, and since the confidence intervals barely include the null, I believe that a more suitable interpretation of those data is that there is a moderately strong association though the estimate of that association is very imprecise.2 Indeed, the confidence interval in any one category of anger is really irrelevant; the important question is whether there is a linear trend in the odds ratios across categories of anger.3 Can the authors report data like those in Table 3 for both intentional and unintentional injuries, and evaluate the ORs for linear trend, to help clarify this issue? Thanks! 1.Bureau of Justice Statistics, National Crime Victimization Survey. Criminal Victimization, 2004. Available at: http://www.rainn.org/docs/statistics/ncvs2004.pdf. Accessed January 29, 2006. 2.Poole, C. Low p-values or narrow confidence intervals: which are more durable? Epidemiology 2001; 12:291 – 294. 3.Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998:306-322. Competing interests: None |
|||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH |
TRACK: