Wednesday, August 01, 2007

Why try to trick people?

Marian MacDorman, PhD thinks that the C-section rate in the US is too high. She is entitled to her opinion, but she is not entitled to tout faulty research to claim that C-section increases the risk of neonatal death. A new paper on the reliability of birth certificate data demonstrates AGAIN (if further demonstration were needed) that 2006 article in Birth, Infant and Neonatal Mortality for Primary Cesarean and Vaginal Births to Women with "No Indicated Risk," is so flawed as to be worthless.

MacDorman and colleagues based their paper on information that is known to be completely unreliable. Not surprisingly, their conclusions are completely unreliable, and almost certainly the opposite of the truth. MacDorman et al. calculated neonatal mortality for vaginal deliveries and C-section of women who had no risk factors on the birth certificate. Yet a large amount of previously published research had shown that birth certificates are often filled out wrong. Specifically, information on complications, risk factors and bad outcomes are usually missing. A new paper in the American Journal of Obstetrics and Gynecology confirms the unreliability of birth certificate data.

Accuracy of birth certificate data by risk factors and outcomes: analysis of data from New Jersey compared birth certificates to the actual records of each woman. Table 2 provides information on the true positive rate of various conditions. The true positive rate, or sensitivity, is defined as the proportion of cases that had a given condition that were identified correctly as having that condition in the certificate. Let's look at risk factors and complications and how often they appear on the birth certificate:

diabetes 42%
pregnancy associated high blood pressure 20%
pre-existing high blood pressure 18%
abruption 29%
placenta previa 40%
cord prolapse 21%
fetal distress 33%

Indeed, there was not a single risk factor, pre-existing condition or complication that was noted even 50% of the time. It is also important to understand that these errors are non-random. All records are not equally likely to be affected. Only the records of women who had risk factors, pre-existing conditions and complications are affected.

Birth should withdraw the MacDorman study since it is a piece of junk. They should insist that MacDorman et al. go back and pull the records of all birth certificates that had "no indicated risk" and remove those that failed to include existing risk factors, pre-existing conditions or complications. Then they can publish the corrected result. Almost certainly, when correctly done the study will show that C-sections for "no indicated risk" have the same neonatal mortality rate as vaginal deliveries and possibly even a lower neonatal mortality rate. That's not going to happen, though, because that is not the conclusion that Birth and MacDorman et al. want to hear. They'd rather publish known erroneous conclusions that support their philosophy than true conclusions that don't.

0 Old Comments: