NHAMCS Publications Avoid Disturbing Misrepresentations on Causes of Injury

face of a prisoner

In the early 1990s, claims that domestic violence is the leading cause of women’s visits to hospital emergency departments became prevalent in U.S. public discourse. Beginning in 1992, the U.S. Centers for Disease Control, an expert agency under the U.S. Department of Health and Human Services, has conducted an annual survey of visits to hospital emergency departments. The survey is called the National Hospital Ambulatory Medical Care Survey (NHAMCS ). NHAMCS provides data on causes of injury. NHAMCS 1992 data, as well as much subsequent data, indicated that domestic violence is far from the leading cause of injury to women. Official publications summarizing NHAMCS data, however, largely avoided showing causes of injury by sex and informatively ranking causes of injury.

The NHAMCS emergency department summary for 1992 did not clearly present causes of injury visits. This report was published on March 2, 1994. It stated:

‘‘Other accidents’’ was the most frequently recorded cause of injury and represented 35.8 percent of visits in which a cause was reported. Accidental falls (26.6 percent) and motor vehicle accidents (14.3 percent) were also prominent on the list.^

The category “Other accidents” included a large number of accidents reported in more detail in sub-categories such as accidental injury from cutting or piercing instruments. The largest detailed subcategories absorbed in the heading “other accidents” accounted for about twenty times as many injury-related visits as other categories explicitly listed. Thus the published summary’s cause categories poorly presented the leading causes of injury. Despite the public prominence of claims that domestic violence is the leading cause of injury to women, the published summary did not differentiate causes of injury by sex.

A subsequent NHAMCS report on causes of 1992 injury-related visits to hospital emergency departments tabulated causes of injury by sex, but did not comment on them. This report was published on February 1, 1995. It observed that the cause categories in the earlier summary report “might be combined in a better way.”^ It reported “the top five causes of injury” via six categories:

  1. Accidental falls
  2. Motor vehicle accidents
  3. Struck accidentally
  4. Cut by sharp objects {accidentally}
  5. Violence
  6. Other {including non-coded} ^

These categories were tabulated by sex for injury visit rates per 1,000 persons. The table showed accidental falls to be the leading cause of injury-related visits for females. Accidental falls accounted for about twenty times as many injury-related visits for females as did violence, which is a superset of domestic violence. Despite the public prominence of claims that domestic violence is the leading cause of injury to women, the report did not discuss leading causes of injury by sex.

A 1998 NHAMCS report on causes of injury further obscured causes of injury by sex. This report summarized NHAMCS data for 1992-1995. It reported injury-related visits in cause categories too broad to rank informatively leading causes of injury. Specifically, it reported provided injury-related visits by sex in the broad categories

  1. unintentional
  2. intentional {with sub-categories for “self-inflicted,” “assault,” and “other violence”}
  3. undetermined intent
  4. cause missing ^

The category “intentional” is a superset of injury purposely inflicted by another person (“violence”), which in turn is a superset of domestic violence. Injury-related visits from intentional injuries to females amounted to 5.8% of injury-related visits from unintentional injuries. While these broad categories of injury do not support a meaningful ranking of the leading cause of injury, they indicate that domestic violence accounts for a relatively small share of females’ injury-related visits to hospital emergency departments.

The 1998 NHAMCS report also reported injury-related visits in cause categories too narrow to rank informatively leading causes of injury. The report listed by sex “the 20 most frequent external causes of injury,” as defined by “3-digit ICD-9-CM E-code.”^ The E-codes separate intentional violence from accidental injuries. However, injures reported in the NHAMCS covered about 170 3-digit E-codes. The roughly 150 E-codes falling below the top 20 were categorized in the table as “all other causes.” This category encompassed 69% more injury visits than the table-leading E-code, which was “{accidental} Other and unspecified falls.” Compared to the earlier published NHAMCS table of the top five causes of injury (see above), this table much less meaningfully communicated leading causes of injury.

The 1998 NHAMCS publication also created a new injury classification scheme that obscured violence as a cause of injury. This new scheme combined E-codes for accidents and intentional violence into classes of “mechanism or cause.” The new scheme swept about half of assaults into the broad mechanism classes “other mechanism,” “not elsewhere classified,” and “mechanism not specified.” Using the new scheme, the report tabulated injuries by age-sex groups.^ The first injury-cause tables in this publication use the new scheme. They are the only tables concerning cause of injury that report age-sex groups. Because of the new scheme, that table provided little information to evaluate the claim that domestic violence is the leading cause of injury to women ages 15 to 44.

The 1998 NHAMCS report implicitly referenced claims that domestic violence is the leading cause of injury to women. The 1998 report stated:

It is very important {emphasis added} to note that the ICD-9-CM coding guidelines indicate that if the Patient Record form did not indicate the intent of the injury (for example, the record stated “patient fell down the stairs”), then the intent defaults to unintentional.^

The report did not mention that the survey instrument included explicitly the cause category “injury undetermined whether accidentally or purposely inflicted.” Moreover, the report printed the statement above in emphasized type. A common-sense understanding of accidents is that they are not associated with intent. The statement “the patient fell off the wall” does not literally exclude the possibility that the patient was pushed. On the other hand, if the patient was pushed, that would be a cause of the injury, with E-code 968.1, “Pushing from a high place.” The report further explained:

the estimates in this report for intentional injuries are probably underestimates, and the estimates for unintentional injuries are probably overestimates. In addition, patients presenting to emergency departments with injuries caused by domestic violence are less likely to volunteer the information to medical staff and may report the injuries as unintentional.{two endnotes}

The quoted sentence beginning with the rather awkward connecting phrase “In addition” suggests a plausible explanation for the report’s peculiar tabulation schemes and speculations. Claims that domestic violence is the leading cause of injury to women had become prevalent across public discourses. The 1998 NHAMCS report supported rationalization of those claims. The report did not correct with relevant NHAMCS findings statements declaring an enormous magnitude of (men’s) (criminal) domestic violence against women.

Summaries of NHAMCS data subsequently avoided the issue. Since 1995, annual general summaries of NHAMCS have included a table of injury causes categorized informatively, but not separated by sex. The 1998 NHAMCS report on injury-related visits emphasized that intentional juries were probably underestimated. It appended to several tables on causes of injury the footnote, “Figures for intentional injuries may be underestimated due to underreporting.”^ Subsequent annual summaries, in contrast, did not include this footnoted speculation. In fact, the annual summary report published in 1999 described a case of over-reporting of intentional injuries:

Discrepancies may arise in respondent interpretation of intent. For example, in some cases, hospital staff checked the “assault” category for dog bite injuries. However, dog bites are an unintentional injury based on the ICD-9-CM E-codes.^

This text was included the next four annual NHAMCS emergency department summary publications. However, after 1998, statistics on injury-related visits received much less attention in NHAMCS publications. After the 1998 NHAMCS report, no subsequent such publication focused on injury-related visits to hospital emergency departments.

Detailed analysis of NHAMCS summary publications shows that NHAMCS data did not factually inform prevalent, highly damaging public claims that domestic violence is the leading cause of injury to women. A public use data tape with the 1992 NHAMCS data was available by September, 1994. After that date, anyone could have produced tabulations of 1992 NHAMCS data showing that accidental falls accounted for more than five times as many of women’s injury-related visits to hospital emergency departments as did interpersonal violence (a superset of domestic violence). The U.S. National Center for Health Statistics, which produced the NHAMCS reports discussed above, obscured such facts in its publications on causes of injury. Clarifying relevant facts probably would not have raised the perceived public value of NHAMCS data and may have significantly undermined political support for the data collection. Concrete support for truth in real, difficult circumstances cannot be taken for granted even from highly credible sources.

Leave a comment (will be included in public domain license)

Your email address will not be published. Required fields are marked *