Lancet editor Richard Horton wrote in an editorial accompanying the survey that more household clusters would have improved the precision of the report, "but at an enormous and unacceptable risk to the team of interviewers."
"This remarkable piece of work represents the efforts of a courageous team of scientists," he wrote.
To conduct the survey, investigators visited 33 neighborhoods spread evenly across the country in September, randomly selecting clusters of 30 households to sample. Of the 988 households visited, 808, consisting of 7,868 people, agreed to participate. Each household was asked how many people lived in the home and how many births and deaths there had been since January 2002.
The scientists then compared death rates in the 15 months before the invasion with those that occurred during the 18 months after the attack and adjusted those numbers to account for the different time periods.
Even though the sample size appears small, this type of survey is considered accurate and acceptable by scientists and was used to calculate war deaths in Kosovo in the late 1990s.
The investigators worked in teams of three. Five of the six Iraqi interviewers were doctors and all six were fluent in English and Arabic.
In the households reporting deaths, the person who died had to be living there at the time of the death and for more than two months before to be counted. In an attempt at firmer confirmation, the interviewers asked for death certificates in 78 households and were provided them 63 times.
There were 46 deaths in the surveyed households before the war. After the invasion, there were 142 deaths. That is an increase from 5 deaths per 1,000 people per year to 12.3 per 1,000 people per year - more than double.
However, more than a third of the post-invasion deaths were reported in one cluster of households in the city Fallujah, where fighting has been most intense recently. Because the fighting was so severe there, the numbers from that location may have exaggerated the overall picture.
When the researchers recalculated the effect of the war without the statistics from Fallujah, the deaths end up at 7.9 per 1,000 people per year - still 1.5 times higher than before the war.
Even with Fallujah factored out, the survey "indicates that the death toll associated with the invasion and occupation of Iraq is more likely than not about 100,000 people, and may be much higher," the report said.
The most common causes of death before the invasion of Iraq were heart attacks, strokes and other chronic diseases. However, after the invasion, violence was recorded as the primary cause of death and was mainly attributed to coalition forces - with about 95 percent of those deaths caused by bombs or fire from helicopter gunships.
Violent deaths - defined as those brought about by the intentional act of others - were reported in 15 of the 33 clusters. The chances of a violent death were 58 times higher after the invasion than before it, the researchers said.
Twelve of the 73 violent deaths were not attributed to coalition forces. The researchers said 28 children were killed by coalition forces in the survey households. Infant mortality rose from 29 deaths per 1,000 live births before the war to 57 deaths per 1,000 afterward.
