by revere, cross-posted from Effect Measure

The National Health Interview Survey (NHIS) has been conducted since 1957 and is one of the main instruments to get a glimpse at the health of the US population. The NHIS is a “multistage area probability design,” or what many call cluster sampling. The idea is to first sample geographic areas in all 50 sates and the District of Columbia, where the area might be a county, a small adjoining group of counties in sparsely populated places or a metropolitan area where population is dense. The list of areas to be sampled has about 1900 entries and 428 are drawn at random, although all states are sampled. So that’s the first stage. Within these primary sampling units subsamples of either 8, 12 or 16 addresses are then drawn or a sample of 4 housing units built after the year 2000. The sample isn’t a straight random sample of the population as Blacks, Hispanics and Asians are oversampled via a screening procedure or by including areas known by the previous census to have higher numbers of these demographic groups. You can read more details at the NHIS site. Interviewing of sampled households goes on throughout the year to get a representative sample of adult, non-institutionalized resident of the US. Participation is voluntary, but response rates exceed 90%, which is pretty amazing.

This week CDC released results about rates of injury episodes collected via NHIS interviews from 2004 to 2007. An injury episode was defined to be “physical damage to the body from external causes resulting from traumatic events, which can include intentional or unintentional injuries.” The question asked was whether there was a nonfatal, medically attended injury in the 5 weeks preceding the interview. Here “external cause” refers to how the injury took place, although CDC uses (to this epidemiologist’s eyes) at least one strange category as “cause”: cut or pierce, which is an effect, not a cause. I’m told this has to do with coding of external causes in discharge diagnoses, but I don’t really know. In any event, here are the results for the “causes” of “falls,” “overexertion,” “struck,” “transportation,” and “cut or pierced”:

Injury quickstat.jpg

Source: CDC, MMWR

Definitions: “Overexertion” denotes excessive physical exercise or strenuous movements in recreational or other activities; “struck” denotes being struck by or against an object or person; “transportation” denotes trauma involving motor vehicles, bicycles, motorcycles, pedestrians, trains, boats, and airplanes; and “cut or pierced” denotes being cut or pierced by instruments or objects.

We see immediately that the big category here is “falls,” accounting for almost 40% of the total, double any other cause. This was true for both males and females, but the rate for females was 17% than for males. On the other hand, females had a 35% lower rate of being struck and a 50% lower rate of being cut or pierced (as an injury, not as adornment). But falls is the big one.

The single category of falls hides a complicated event. I’m no injury expert, but I know there are different kinds of falls, some on the same level (tripping on a throw rug or toy) or from different levels (down steps, off platform or ladders) and affect different age groups differently, with the elderly being at special risk. It’s my impression that the rate of falls in the elderly has been increasing and I’m not sure if it’s know why, but one can speculate that, on the one hand, it’s because the elderly are in better shape and more active, hence more at risk of falls, or that multiple medications for blood pressure or anti-depressants might be involved. Certainly bad weather and poor vision are other risk factors. The cost in hospitalizations and emergency room visits is considerable, with older women being particularly important in this regard. And according to one source I consulted, grandma is more likely to fall on a level surface in her home rather than on the stairs, although it is helping grandma up and down stairs that the public seems to focus on.

Injury is a small specialty in epidemiology but an interesting one. Most “accidents” aren’t freak events but the result of something both foreseeable and preventable. Or to use a hoary clichĂ© from the injury community, “Injuries are not accidents.”