Deltoid

Point Blank, by Gary Kleck, pg 165, citing a study by Wilson and Sherman, 1961:

"At least one medical study compared very similar sets of wounds ('all were penetrating wounds of the abdomen'), and found that the mortality rate in pistol wounds was 16.8%, while the rate was 14.3% for ice pick wounds and 13.3% for butcher knife wounds.

The study is in Annals of Surgery Vol 153 pp 639-649 "Civilian Penetrating Wounds of the Abdomen" by Wilson and Sherman. It covers stab (5% mortality) and gun shot wounds (17% mortality) to the abdomen.

The numbers Kleck quotes above come from Table 7 of the article which contains mortality data by weapon. The implication seems to be that "knives are almost as deadly as guns". This is extremely misleading.

There are two basic questions to be answered:

1. Exactly what was measured?

2. Is the result statistically significant?

(1) The data is from 452 admissions with abdominal wounds to a hospital in Memphis, Tennessee over the period 1948-1959.

(1a) People who died before reaching hospital are NOT counted. In the discussion following the paper it is stated that "the preponderance of stab wounds is more apparent than real because a significant percentage of patients wounded by gunshot die before reaching the hospital.", so this will make the mortality rate for gunshot wounds appear to be less.

(1b) The wounds include self-inflicted and accidental cases. Someone attempting suicide with a gun will probably aim at the head, but a a would-be knife suicide may well attempt disembowelment.

(1c) Mortality rates for wounds to other parts of the body may well be very different. For example, a low velocity weapon like a knife is far less likely to penetrate a skull than a high velocity projectile.

(1d) The distribution of wounds is different for knife assaults and gun assaults, since victims of knife assaults have more chance to dodge and block.

(1e) Medical treatment has improved since 1948. More recent results on abdominal wound mortality (Annals of Surgery 179 pp 639) show that stab wounds are 1% lethal and gun shot wounds are 13% lethal.

(1f) The weapon used was known for only some of the cases. The mortality rate for gunshot wounds where the type of gun was unknown was 29%, so this made the mortalities for each type of gun appear to be lower than they really were.

(2) The 13.3% death rate for butcher knife wounds is based on a mere 15 cases. This is far too few to give a meaningful mortality rate. The death rate for rifle wounds was 7.7% (based on only 26 cases). Do you think rifles are half as lethal as handguns?

I have calculated 95% confidence intervals for each of the weapons in the paper. Here are the results:

Weapon     Cases    Deaths   % Deaths    95% conf for mortality rate
Shotgun       49        10       20.4    11%-34%
Pistol       101        17       16.8    11%-25%
Ice Pick      14         2       14.3    4%-40%
Butcher Knife 15         2       13.3    4%-38%
Rifle         26         2        7.7    2%-24%
Switch-blade
knife         17         1        5.9    1%-27%
Pocket knife  44         0          0    0%-8%
Unknown GSW   14         4       28.6    12%-55%
Other stab   172         9        5.2    3%-10%

All GSW      190        33       17.4    13%-23%
All stab     262        14        5.3    3%-9%

We see that mortalities for each pointed weapon are not significantly different from mortalities for all pointed weapons, but that mortalities for stab wounds are significantly less than mortalities from gun shots.

95% confidence intervals for mortalities calculated from (Annals of Surgery 179 pp 639) are 1%-2% for abdominal stab wounds, and 11%-15% for abdominal gun shot wounds.