Where do those abortion estimates come from?

The International Organizations Research Group has published a paper by Donna Harrison, M.D. focusing on the how the World Health Organization comes up with its statistics on worldwide abortion and maternal morbidity along with the nonsensical manner in which the terms "safe abortion" and "unsafe abortion" are defined. Regarding "safe" and "unsafe" abortions:

According to WHO, induced abortion is defined as “the voluntary termination of pregnancy.”8 WHO defines unsafe abortion as “a procedure for terminating an unwanted pregnancy either by persons lacking the necessary skills or in an environment lacking the minimal medical standards or both.”9 Subsequent WHO documents, such as a 2007 article co-sponsored by WHO, allow for a purely legal definition of unsafe abortion as “abortions in countries with restrictive abortion laws.”10

According to the article, any induced abortion, even under the most medically pristine conditions, performed in a country where abortion is “illegal” is deemed “unsafe.” Similarly, any induced abortion performed in a country where abortion is legal, regardless of the subsequent morbidity and mortality which follows, is considered “safe.”

Harrison then shares how researchers who come up with estimates for pregnancy-related deaths must be "committed to adjusting the data" aka making it up to fit your preferences.

At a session entitled “Monitoring MDG 5: Innovations in Measuring Maternal Mortality” on October 19, 2007, Johns Hopkins University’s Dr. Cindy Stanton said, “To participate in interpretation of pregnancy related deaths requires that one be committed to ‘adjust the data.’” Dr. Stanton, a WHO researcher, explained that adjusting data means “eyeballing it to see if it makes sense from what we expect.” This adjusting is especially necessary with “pregnancy related deaths,” she continued, and said, “We adjust the number of births or the number of deaths and we don’t change the number of pregnancy related mortality.” “In some areas,” Stanton went on, “We make huge adjustments to make the numbers turn out right. More than fifty percent of some numbers are ‘adjusted.’”