Based on new and re-analyzed data, it seems that maternal mortality rates jumped up almost 27 percent between 2000 and 2014 in the U.S, according to a new study published in the journal Obstetrics and Gynecology. That trend moved opposite to the trend of declining maternal mortality rates in the rest of the world and was directly opposed to the U.N. Millennium Development Goal of a 75 percent reduction in maternal mortality by 2015.

This also means the U.S. has one of the worst maternal mortality rates among developed countries.

In 2000, about 18.8 women per 100,000 live births died in relation to pregnancy and delivery, while childbirth complications killed about 23.8 women per 100,000 live births by 2014.

The death rates in Texas and California fell outside the normal trends seen in every other state, possibly due to unique demographic characteristics or changing health circumstances. Texas’ maternal death rate (especially between 2011 and 2014) increased much more sharply than the rest of the country’s, while California’s appeared to be the only one that decreased.

Efforts to change those numbers could depend on understanding those numbers. Researchers looked at 48 states and Washington, D.C., while separately analyzing Texas and California.

The U.S.’s maternal mortality rates were poorly understood for several years in the 2000s, the researchers pointed out. By looking at the raw number of maternal deaths as reported, it would appear that it more than doubled from 9.8 per 1000,000 in 2000 to 21.5 per 100,000 in 2014.

But there is a different story in the numbers, because of a change in recording previous or current pregnancy status on death certificates (and an uneven adoption of the changing protocols among various states). Starting in 2003, states began to update the information recorded on death certificates to match the maternal death info used for international comparison by the World Health Organization.

The parameters include recording whether a deceased woman has been pregnant in the last 42 days or in the last year and whether or not her death was related to pregnancy or childbirth. But different states adopted the new recording standards at different times. It was only by 2014 that most states’ (all except California, Colorado, Massachusetts, Virginia and West Virginia) death certificates conformed to the new updates.

The researchers explained, “Due to difficulties in disentangling these effects, the U.S. has not published an official maternal mortality rate since 2007.” So these researchers undertook an effort to understand these trends themselves.

They couldn’t extrapolate any statistically significant information from looking at states individually (using their individual standards), because the absolute number of maternal deaths in the U.S. is so low—just 396 in 2000 and 856 in 2014. So they developed a formula to match certain states with similar characteristics and dissimilar death certificate revision dates.

“The correction factor was developed to adjust unrevised data to be comparable with revised data and was computed for a group of 24 states and Washington, D.C., that, before revision, did not have a pregnancy question,” they explained.

The correction factor was the “sum of the number of maternal deaths in each state for two years following the revision date divided by the sum of the number of maternal deaths in each state for the two years preceding the revision date.”

By analyzing the death rates differently and in groups, they were able to come up with a number that is more comparable to the measurement used in other parts of the world and mostly followed the previously mentioned 18.8 per 100,000 to 23.8 per 100,000 trend that the researchers concluded was the nationwide change.

However, Texas and California have big enough populations to analyze individually—and their results seemed completely different from the rest of the country’s.

In Texas, adjusting for revised questions, the maternal death rate seemed to follow the rest of the country between 2000 and 2010, increasing from about 17.7 to about 18.6. But between 2011 and 2014, the maternal death rate exploded to about 33 maternal deaths per 100,000 live births in 2011 and 35.8 in 2014.

In California, the rate actually decreased. In 2003, its maternal death rate was about 21.5. By 2014, it was at 15.1.

The researchers speculated that the sudden uptick in maternal deaths in Texas was likely not a result of faulty statistical analysis, but it could be attributed to a changing health environment, partially related to the closing of women’s health clinics. But the researchers also insisted there must be some other explanation to account for such a sudden doubling of maternal deaths absent some kind of catastrophe. They called for further research in the state.

Conversely, California’s decline can likely be attributed to its California Maternal Quality Care Collaborative effort to reduce such mortality, such as by focusing on “evidence-based” ways to reduce hemorrhaging and preeclampsia, two of significant causes of maternal death.

The researchers called for a nationwide effort to follow California’s maternal health trends and to report them accurately to international monitoring agencies (the inability to do so they called an “international embarrassment”). That would mean increasing funding to statistical agencies and “redoubling” efforts to minimize maternal deaths.