Working with California Watch, I first broke the story on a rise in maternal death rates in February. After consulting with California Watch editors Robert Salladay and Mark Katches, I decided to delve deeper into the area of cesarean sections, which have been rising during the past several years.
I started by looking at what information the state already had. The most interesting thing I found was that California’s Office of Statewide Health Planning and Development had done a statistical analysis comparing C-section rates at California hospitals and noted that the wide variation in rates could indicate that some hospitals were performing unnecessary surgeries. To refine this point, the state focused on those C-sections most likely to be unnecessary.
The state data excluded deliveries in which C-sections might be medically justified: instances in which the woman had a prior C-section, twins, breech babies and babies delivered before term (which covered babies induced early due to diabetes, heart disease, eclampsia and other complications). The state data also excluded births that resulted in the death of the baby.
Finally, it adjusted these rates for maternal age because older mothers are more likely to require a C-section. The patients remaining after making all these exclusions were women who were most unlikely to need a C-section for medical reasons. This “low-risk” rate formed the foundation of my analysis.
Before delving into the data, I asked two experts in the field, Debra Bingham, president-elect of Lamaze International, and Dr. Elliott Main, principal investigator for the California Maternal Quality Care Collaborative, how to best analyze this variation between hospitals. They suggested I look for correlations between the low-risk C-section rate and other factors, such as type of insurance, and hospital quality measures. I also spoke with experts at the Robert Wood Johnson Foundation, Kaiser Family Foundation and California HealthCare Foundation.
With the help of California Watch intern Timothy Sandoval, I combined the numbers from the state’s C-section analysis, from 2005 to 2007, into a single Excel spreadsheet. Entering data by hand, we then added information on hospital characteristics and finances. Most of this information came from the state, but I also used federal data from patient surveys about hospitals, as well as data from the Dartmouth Atlas of Health Care, an academic research organization. I also made public records requests, which resulted in disclosure of Medicare fees paid for C-sections.
California has about 430 hospitals and medical centers. Of that total, nearly 260 hospitals reported birthing statistics to the state during the timeframe analyzed.
After looking over this spreadsheet I excluded from the analysis five hospitals with significant missing data: Valleycare Medical Center in Alameda County; Kaiser Permanente Antioch Medical Center in Contra Costa County; Selma Community Hospital in Fresno County; Martin Luther King Jr. Hospital in Los Angeles County; and Rideout Memorial Hospital in Yuba County. Martin Luther King Jr. Hospital shut down in 2007, leaving significant gaps in its data. The four other hospitals were missing C-section data for most of the three years (2005-07).
I then began looking for statistical correlations between the low-risk C-section rates and various hospital characteristics. To do this, I used the Pearson’s correlation function in Excel, which shows if there is a linear relationship between two columns. The formula yields a correlation coefficient between -1 and 1. A perfectly linear correlation is indicated by a coefficient of + or -1, while 0 indicates no correlation. So, for instance, if hospitals with higher low-income patients (i.e. patients utilizing Medicaid and indigent assistance)rates always had higher C-section rates, the Pearson’s coefficient would be close to 1.
But, as it turned out, the percentage of a hospital’s low-income patients was not correlated with C-section rates. The formula yielded a correlation coefficient of 0.1. The graph comparing C-section rates with low-income patient care rates looks like a random cloud of dots, rather than a straight line.
Those dots tightened into something resembling an oval when I compared C-section rates related to a hospital’s exclusive breast-feeding rate: The formula produced a coefficient of -0.3, a weak inverse correlation.
Next, I wanted to see if either for-profit or nonprofit hospitals were more likely to perform C-sections, but because one column contained words (e.g., nonprofit) rather than numbers, I couldn’t run Pearson’s formula. Instead, I calculated the average low-risk C-section rate for nonprofits and compared that to the for-profit average. I included all hospitals that don’t earn money for investors or private owners in the nonprofit set (that is, all hospitals run by nonprofits, charities, tax districts, cities and counties). This division left me with 48 for-profit and 205 nonprofit hospitals.
The average low-risk C-section rate was higher at for-profit hospitals by 3.7 percentage points. This difference was found to be statistically significant using a two-tailed T-test: (t = 3.72, p < 0.001, d.f. = 251).
I shared this result with Gene Declercq, professor of community health sciences at the Boston University School of Public Health. Declercq, who is nationally recognized for his work with birth statistics, asked to see the spreadsheets. After looking over the work, Declercq said the methodology was fair and complied with standard practice. Declercq said I should weight the averages by number of births, so that small hospitals wouldn’t have an outsized importance in my analysis.
To weight the averages, I multiplied low-risk C-section rates by the average number of births at each hospital from 2005 to 2007, and then divided by the average number of births per hospital. After weighting the averages for number of deliveries per hospital, the difference was 2.78 percentage points, meaning that a pregnant woman walking into a for-profit hospital is 17 percent more likely to receive a C-section than a pregnant woman walking into a nonprofit hospital.
Declercq also pointed out that, although I had included county demographic information, a report with better access to data would have controlled for demographic factors (such as higher rates of poverty and obesity in some areas) specific to the particular C-section patients at each hospital. Because the 48 for-profit and 205 nonprofit hospitals are scattered throughout the state, it’s probably safe to assume that for-profit hospitals are no more likely than nonprofit medical centers to be in areas of high obesity. But a more thorough analysis would quantify and exclude these confounding demographic factors. We decided against doing that and simply focused on county demographic information.
California Watch’s Agustin Armendariz converted the most useful numbers in the spreadsheet to a visual format accessible on the web. California Watch’s Chase Davis checked the math and statistical calculations. California Watch copy editor Austin Fast also double-checked the math.
Most of the work I did for this story was not statistical, but instead comprised old-fashioned telephone and shoe-leather reporting, visiting hospitals at both ends of the spectrum and interviewing researchers, hospital administrators, doctors and nurses. I put out a call for women who believed they had received an unnecessary C-section through online groups, doula organizations, childbirth educators and the advocacy group International Cesarean Awareness Network. The two women quoted in the story were both associated with this last group.
On the question of the profits hospitals make from C-sections, I relied on a report [PDF] from the California Pacific Business Group on Health. I checked the estimates in that report in off-the-record interviews with insurance company representatives. Insurance companies declined to openly reveal the amounts they paid to specific hospitals for C-sections and vaginal deliveries, but they provided representative reimbursement amounts without naming hospitals.
Each person and hospital mentioned in the story was given the opportunity to comment.
Nathanael Johnson is a freelance contributor to California Watch.
Comments
via Twitter