New paper on C-sections is misleading and leaves out important data

Male doctor giving thumbs down

A new paper in the journal Health Affairs is receiving a great deal of media attention. Although the paper provides valuable data, the authors dramatically overstate the conclusions and leave out critical information.

The paper is Cesarean Delivery Rates Vary Tenfold Among US Hospitals; Reducing Variation May Address Quality And Cost Issues by Kozhimannil et al. According to the authors:

Working with 2009 data from 593 US hospitals nationwide, we found that cesarean rates varied tenfold across hospitals, from 7.1 percent to 69.9 percent. Even for women with lower-risk pregnancies, in which more limited variation might be expected, cesarean rates varied fifteenfold, from 2.4 percent to 36.5 percent. Thus, vast differences in practice patterns are likely to be driving the costly overuse of cesarean delivery in many US hospitals.

Not exactly. To see how the authors overstate their case, it helps to looks at charts that they created.

US C-sections 2009

Yes, it is true that the rate at the hospital that did the greatest proportion of C-section is 10 times higher than the rate at the hospital that did the lowest proportion of C-sections, but a glance shows that both hospitals are outliers. Therefore, that comparison is essentially useless. A far more valuable statistic is the interquartile range, the difference between the 25-75 percentiles. As the authors acknowledge in a subsequent table, the mean C-section rate in 2009 was 32.8 with an interquartile range of 9.4. So fully half of the hospitals had C-section rates in the range of 23.4%-42.2%. That’s still an appreciable difference (double), but very far from the 10 fold difference touted by the authors. In fact, more than 90% of hospitals had C-section rates between 21%-44%.

The same thing applies to the analysis of C-section rates in low risk women.

US low risk C-sections 2009

The authors report that the C-section rate for low risk women varies 15 fold among hospitals, but that is misleading. As the authors acknowledge in the subsequent table, the mean C-section rate for low risk women in 2009 was 12 with an interquartile range of 4.9. Fully half of the hospitals had low risk C-section rates ranging from 7.1%-16.9%. Again the difference is appreciable (slightly more than double), but a very far cry from a 15 fold difference. Nearly 90% of hospitals had C-section rates for low risk women between 6%-19%.

So the national variation in C-section rates is far less than the authors claim. Moreover, the authors commit the same error as do many natural childbirth advocates; they focus on process as opposed to outcome. We shouldn’t be looking for an ideal average C-section rate. We should be looking for the C-section rate that produces the best outcomes. How does the perinatal mortality rate compare between hospitals with low C-section rates and high C-section rates? The authors don’t know because they never looked. Indeed, the underlying (and totally unjustified) assumption that permeates the entire study is that there is no appreciable difference in mortality rates between various hospitals and that, therefore, we can focus in difference in C-section rates.

But perinatal mortality rates do vary appreciably among hospitals and it is critical to include this data. What if the mortality data showed that hospitals with C-section rates below 25% have higher perinatal mortality rates than hospitals with higher C-section rates. If that were the case, the hospitals with lower rates should be chastised, not held up as a model for an ideal, achievable C-section rate.

The authors conclude their paper with the following:

Although some variation would reasonably be expected given differences in patient populations, the scale of the variation in hospital cesarean delivery rates—most notably, a fifteenfold variation among the lower-risk subgroup— indicated a wide range in obstetric care practice patterns across hospitals and signaled potential quality concerns.

But as we have seen, the real variation among hospitals is much smaller making it much less likely that differences are due to practice patterns. Most importantly, the authors are not in a position to assess quality concerns unless and until they look at outcomes, and privilege them above process.

  • Mrs. W
  • K M Johnston

    I’m not a scientist, but I would have done a bar chart for both of those which would have been a nice bell curve, immediately making obvious both outliers and median. Why pick this arbitrary chart type where I have to stare at it to figure out wat it says?

    • The Bofa on the Sofa

      As an old friend used to say, “If you don’t have anything to say, say it in color.”

  • Lisa from NY


    ” A San Francisco study of 8,500 first-time moms who delivered full-term, head-down babies found that the older they were, the longer they labored. Their cervixes dilated more slowly, and it took longer for them to push their babies out. Not surprisingly, then, the older they were, the more likely they were to receive oxytocin (frequently referred to by one of its brand names, Pitocin) to ramp up their labor. But it’s not like your uterus suddenly heads south when you hit 35. Here’s a cheery thought: The study found that the chance that your uterus has lost its oomph actually begins to increase when you’re in your early 20s — long before you begin to entertain thoughts of Botox or bifocals.”

  • Lisa from NY

    It also doesn’t address the advanced maternal age of mothers today, and increases in diabetes, herpes, HIV…

    And studies do show that many women giving birth first time after 40 CAN’T push the baby out and need a C.

    I am sick and tired of my 50 yr old pregnant neighbor telling me that she would be putting her baby’s life at risk if she would deliver in a hospital (first was via C at age 48) and taking Chinese herbs will solve all problems. (?)

  • Dr Kitty

    This study is a bit like saying “GP A had 75 patients die last year and GP B had 17 patients die-GP A is obviously a terrible Dr!”

    Without noting that GP A has a list size looks after 8 nursing homes and works in the inner city, while GP B works in a college town and mainly cares for students and college staff with young children.

    It doesn’t actually tell you whether if the GPs switched and looked after each other’s patients whether the number of deaths would increase or decrease.
    Perhaps if GP A looked after the college town only 5 patients would have died, perhaps if GP B looked after the nursing homes 72 patients would have died.

    Who can say?
    This study can’t.

  • SkepticalGuest

    I’m guessing that some of the outlier hospitals with higher c-section rates are hospitals that specialize in taking high-risk patients. We have a local university hospital with a super-high c-section rate, but there are people who come from other STATES to deliver there if there are serious complications, and certainly the same phenomenon happens within the city as well. Wonder if the outlier low-c-section rate places are smaller, ill-equipped hospitals that only low-risk–or not in the know–women go to. In fact, one has to wonder if the outlier low-CS hospitals has vaginal births that would’ve been better off as C-sections but couldn’t be due to lack of staff. I’m always surprised how many hospitals, even in large metropolitan areas, don’t even have 24/7 anesthesiology coverage.

    • Charlotte

      It’s the same way in my area. The hospitals with the highest c-section rates are the ones with NICU’s considered to be among the very best in the entire country. Their stats include tons of c-section moms who were sent there from hundreds of miles away because their doctors knew it was the best place for their high-risk delivery and medically fragile baby.

  • quadrophenic

    OT: Gina just posted a link to her blog on Facebook about her response to Dr. Amy’s suit. Apparently it just links to the response she filed. So the response has now been filed/served and is available online. I’d love it if someone with a PACER log in or Dr. Amy could post it. My PACER log in is tied to work and I’ll get a nasty email if I use it for something personal.


      GCC’s obvious contributions are funny. I don’t understand the legalese, but the gist is: We feel the case should be dismissed on grounds X, Y, Z.

      • LukesCook

        Please tell me that document wasn’t drafted by an actual attorney. An attorney who filed those papers here would be cruising for a de bonis propriis costs order.

        • auntbea

          What’s that in American?

          • LukesCook

            Don’t speak American, couldn’t say.

        • “SPECIFIC

          Costs de bonis propriis

          Awarded against person acting in a representative capacity.
          Such costs are a penalty for improper conduct and the representing
          person must pay out of his own pocket.”

          Improper conduct by GCC’s counsel?

          • LukesCook

            The various aspersions cast on both Dr Amy and her husband are totally irrelevant to the legal issues raised. Their only purpose is to tarnish their professional reputations and as such are an abuse of court process and of litigation privilege. An attorney who wilfully abused court process here could expect a tongue-lashing from the judge if he was lucky, most likely an unfavorable remark in the written judgment (not good when it comes up when a potential client or employer searches on your name), and a de bonus propriis costs order if the circumstances warranted it. In the most egregious cases the judge might address a letter or copy of his judgment to the relevant professional body for investigation and possibly disciplinary action. Maybe American courts are more tolerant of their time and records being wasted with inappropriate and irrelevant personal slurs, but papers like these wouldn’t be allowed to pass here.

      • LovleAnjel

        They really did argue that Dr. Amy’s mean! IN A LEGAL DOCUMENT. she needs to get herself a new lawyer.

        • I suspect GCC has become painfully aware that all court documents are public and is trying to publicize her personal grievance by including in this response.

      • Charlotte

        Isn’t this her third or fourth lawyer at this point?

      • The Computer Ate My Nym

        Is this for real? It reads like a satire of what GCC would say in a complaint if her lawyer didn’t talk some sense into her.

        • mollyb

          I didn’t know I could use the fact that i have young children and am a student to get out of legal trouble. Sweet! Off to rob a Publix.

      • theNormalDistribution

        Does Dr. Amy actually run this “ask Dr. Amy” site? It looks kind of unprofessional and not at all in keeping with the quality of her other sites.

        • Poogles

          “Does Dr. Amy actually run this “ask Dr. Amy” site?”

          I know it is her site, though I am not 100% sure she does all the running/maintaining of the site or not.

        • auntbea

          She has explained elsewhere that it is just a means to prevent people from barraging her with emailed medical questions all day.

    • Kalacirya

      I put all the new docs up on Scribd.

      They include the memo posted by Anj, affidavits by both of Gina’s past attorneys and one from herself. And a bunch of screenshots from this blog and other places Dr. Amy has written.

      • Thanks, very interesting.

  • anonymous

    Useless data. I work in statistics and Amy’s right. You’ve also got to look at how things are organized. The hospital I have privileges at has a separate maternity department to handle all the NICU instances and they’re technically classed as their own hospital. Would that mean that our hospital, which sees a delivery now and then in the ER rank low and them rank high? I didn’t read the article, since I don’t have a subscription but I don’t see which percentage of these were elective. Both of my children were delivered by cesarean and it was my wire and my choice.

  • AllieFoyle

    This a cruddy study and I think anyone with half a brain could effectively argue that it’s too flawed to be taken seriously, but…no matter how poor it is, it just adds weight to the anti-CS canon. It’s clearly just another study designed and executed to make a c-section look like a public health problem instead of a life-saving, health-preserving way to deliver a baby, but where are the studies on the other side? Where is the real analysis? Why are people only interested in these simplistic explanations for a phenomenon that’s really very complex?

    The scariest part of this paper is the idea that this mass of data points that proves nothing somehow should be used as a basis for tying Medicaid payments to lowering c-section rates. Placing incentives on doctors and hospitals to lower c-section rates will invariably restrict the rights of women who would prefer c-sections and will pressure doctors to pursue vaginal birth in cases where it isn’t in the best interest of women and babies to do so. Some people really are best served by having a cesarean.

    • Mrs. W

      Cesarean is also a mind saver for many women.

      • AllieFoyle


    • theadequatemother

      “The scariest part of this paper is the idea that this mass of data points that proves nothing somehow should be used as a basis for tying Medicaid payments to lowering c-section rates.”

      immediately when I saw this paper used american data, I wondered how much of the variability could be explained by the hodge podge of different insurers with different payments schemes and requirements. But then I was reminded of a study (that I found on the academic OBGYN blog) about the lack of difference between the c/s rates for patients on private insurance vs medicaid.

  • The Computer Ate My Nym

    Mildly off topic, this is interesting:

    They found higher rates of serious fetal/newborn and maternal complications in the VBAC versus the planned repeat C-section group. Maybe the whole VBAC concept should be reconsidered.

    • theadequatemother

      the same was found by perinatal services BC…VBACs had higher morbidity for WOMEN!

      • The Computer Ate My Nym

        In this study, there was a higher morbidity for both woman and babies. So who benefits?

        • Also true of the study comparing homebirth and hospital birth, by the way (with homebirth “winning” by that measure), though that is certainly outweighed, I would agree, by the increased *mortality* risk in the other direction.

          • The Computer Ate My Nym

            Could you give a reference to the study in question?

          • The Computer Ate My Nym

            The paper above is on PLOS Medicine so it’s open access. They found lower rates of major hemorrhage, uterine rupture, and organ damage requiring repair with ERC compared with VBAC, but only major hemorrhage showed a statistically significant difference and there was a non-significant trend towards a greater incidence of wound problems in the ERC group. So, these are major problems that could lead to death, not problems which may be pretty unpleasant but are unlikely to kill such as epidural headaches or minor vaginal tearing. But you’re quite right to point out that lower morbidity may not equal lower mortality and may even be associated with higher mortality. There are plenty of examples in medicine and the secondary endpoints shouldn’t be completely trusted. Fortunately for pregnant women, pregnancy related mortality is low enough that it’s hard to get good statistics on it in the context of a clinical trial.

        • Mrs. W

          Those who are looking to reduce their CS rates benefit…

  • K

    What a weird way to graph those data. Were the authors trying to distract the readers with Rorschach ink blots?

  • LovleAnjel

    Those figures are terrible. It’s a normal distribution – why the funky graph? Are they obfuscating?

    • The Bofa on the Sofa

      I don’t quite understand the graphs they are plotting, but from a data standpoint, when I see outliers like those shown here, my first response is, hmm, they are outliers, I wonder what’s going on there to make them different from the bulk?

      Especially that point at 7.1%. The authors make a big deal out of the factor of 10 between the lowest point and the highest point, but from a data standpoint, the factor of 2 between the lowest point and the second lowest point is pretty curious.

      You note that it looks like a normal distribution, but it’s not quite. The tails don’t actually fall off in a Gaussian fashion. In fact, if you actually count dots, the distribution looks mostly like it has a mean of 32 and a standard deviation of 9, meaning that 95% of the data falls within the range of 14 – 50. However, if that were the case, there should only be like one point outside of the range of 7.5 – 58.5, but there are actually 6. The proper interpretation is that these are pretty extreme cases, and one might wonder what it is about them that leads to this type of distortion?

      There is actually a lot to be learned from something like this, but you actually have to look for it.

    • Kalacirya

      It’s not *that* normally distributed, you’d see it more obviously with a quantile plot, which if I didn’t have a midterm I’d put together for you.

  • The Computer Ate My Nym

    I think that there may be an itty, bitty grain of a point in here. If there is a 2 fold difference in c-section rate between hospitals with comparable populations, that does suggest that there may be differences in practice between OBs in different areas. However…

    1. I agree with the statements about statistical errors that others have eloquently made.

    2. If there is a difference, that does not imply (as the authors appear to assume) that the higher c-section rate is worse. As Dr. Tuteur points out, we don’t know the mortality and morbidity stats at the hospitals examined and it may be that the hospitals with low c-section rates have higher rates of mortality and/or morbidity than those with higher rates.

    3. Transfers. A small hospital might transfer a woman in labor to a larger hospital with more specialist care and therefore get the c-section she may later need off their statistics, making them look “better”, but only because they transfered their high risk patient, appropriately, to some place where they could receive better care.

    4. Their definition of “low risk” was kind of shaky. They included preterm birth, multiples, fetal malpresentation and prior c-section delivery as conditions that make patients “high risk”. All very well, but what about maternal indications? Women with pre-eclampsia, EDS, gestational diabetes, and sickle cell anemia (just to select the first few problems with pregnancy that might lead to c-section that popped into my head) would be counted as “low risk” in this paper, but would be at much higher risk of needing a c-section. I also notice that they didn’t control for fetal size, with or without gestational diabetes.

    5. They’re using billing data. It’s a pain in the butt for epidemiologists, but physicians are notorious for not including every diagnosis in the billing. So the fetus may well have had malpresentation, etc and that just wasn’t recorded on the billing sheet (though hopefully it was in the chart.) Furthermore, attendings at teaching hospitals are worse about including comorbidities than physicians in practice (cynically, probably because their income depends less on wringing every penny out of the insurance companies) so the lack of documentation will be unevenly distributed and may lead to a higher apparent rate of non-indicated c-sections at public and teaching hospitals.

    The bottom line is that if we think specific c-sections are not needed, we need to do large studies randomizing women between c-section and attempted vaginal delivery and see whether one group does better than the other. This isn’t going to happen any time soon because the NIH’s funding has been cut and anyway no one wants to do a study that will possibly end up with some women or babies being harmed, even if the harm is minimal (i.e. had an unnecessary c-section or had to have an emergency c-section after an attempt at continued labor) and even if it is already occurring because we don’t have the data necessary to prevent it. If we really want OB care, we have to pay to get the data and I don’t see anyone wanting to do that in this era of tax cuts above all else and drowning the government in a bathtub.

    • Amy Tuteur, MD

      I agree that the variation in mainstream C-section rates from 23%-42% suggests that practice patterns play a far greater role than they should. That’s why it is so disappointing that the authors undermine their own finding by grossly exaggerating it.

      • TheOtherAlice

        I would also be really interested to know what those ‘outliers’ are doing. How the hell are they so far from the rest of the group, and what are their outcomes like?

      • AllieFoyle

        Yes. Also, why is the takeaway that the hospitals with the high c-section rates are doing something wrong, but no scrutiny is applied to the hospitals with the below-average rates? Maybe those low rates are actually reflective of a rate that is too low and presents a danger to patients. I guess we’ll never know because no one seems to care about that side of the equation.

        • The Bofa on the Sofa

          I’m really, really interested in that hospital with only the 7% rate. That is a much more extreme outlier than the ones at the other end.

          Notice it does not show up as an extreme outlier on the low-risk plot. I wonder what it’s low-risk rate is?

    • AllieFoyle

      There may be a grain of truth there, but they haven’t proven that a) there is actually a troubling disparity after controlling for confounding variables, or b) that such a disparity is reason for concern. Another case of assuming that c-sections are a negative to be avoided instead of recognizing that they may be negative or positive for different individuals in different circumstances.

    • AllieFoyle

      “The bottom line is that if we think specific c-sections are not needed, we need to do large studies randomizing women between c-section and attempted vaginal delivery and see whether one group does better than the other. ”

      You know, I agree on one hand, but also feel that studies like these can only tell us so much that isn’t already known. We know that there are certain risks that go along with TOL; we know that CS carries certain risks. We know the approximate magnitude of these risks. We also know that each delivery represents a different set of factors in terms of maternal and fetal characteristics, patient history, preferences, etc. What’s the advantage of using a blunt instrument–public health level decision-making applied in a one-size-fits-all approach–on a problem that could be much better handled with a more nuanced approach? Each patient should be taken as an individual, given truly informed consent about all options and the risks of TOL and CS, and with decision-making tailored to her specific preferences and best interests.

      • The Computer Ate My Nym

        That’s always the danger of having guidelines: people start using them as absolute standards. It’s never going to be possible to have every last situation covered by guidelines and making them policy is dangerous. Because the next patient may be so far off the standard that the guidelines don’t even make sense for her. But, as far as I know, there hasn’t been a true randomized trial of VBAC versus ERC, for example. I’d like more papers that don’t depend on claims data (a notoriously flaky source) or uncontrolled population studies which are easily confounded. I do think that this just isn’t possible in OB any more and understand that you have to work with what you can, but it’s frustrating.

  • theadequatemother

    Ah, Health Affairs. Some important things to note:

    So first off, this is a policy journal, not a medical journal, and as such the focus is different with an emphasis on dull things like costing.

    The authors are from the realm of health policy. None of the three is a physician. From my increasing involvement with health policy types, it has become apparant that they do not have even the slightest grasp on the clinical realities of patient care. If they did, or if they had included an OB/GYN author or even an MD with a clin epi degree, they would have noted the glaring omission of risk adjustment. Clearly the peer reviewers also had no grasp of the clinical realities of delivering obstetrical care. I would bet the reviewers were other health policy types, economists etc.

    Does it surprise us that different hospitals have different c/s rates? No…and it shouldn’t. In fact I would argue that it is entirely appropriate for different hospitals to have widely varying rates for all the reasons noted below. Hospitals that deal with low risk patients will have low rates. Hospitals that deal with high risk patients will have high rates. That is as it should be. If all hospitals with different risk profiles had the same rates, care would be inappropriate and periantal mortality and morbidity would increase. The assumption that differences in c/s rates are due to different practice patterns should equate to the difference in case risk mix. However, it will be read and interpreted as a difference in OBs using their own convenience to schedule c/s.

    I am gladdened by the move in BC and ON to start using the Robson 10 classification when reporting c/s rates ( This will give more meaningful data and should allow differentiation of c/s in high vs low risk cases.

    • auntbea

      Hey! I work routinely in policy and economics, and the flaws in this paper are obvious to me (as they would be to any of my colleagues.) I don’t think the craptasticness that is this paper can be attributed to the fact that social scientists wrote it.

      • Mrs. W

        Same here. The crappiness blares to me, but I am a bit sensitive when it comes to cesareans.

        This is just laziness pure and simple. Why bother to understand why rates vary when a publication is at stake?

        • The Computer Ate My Nym

          I could forgive this if the conclusion were something like “this is preliminary data and further study is needed to understand why rates vary” rather than an immediate call for lowering c-section rates when we don’t even know whether that is a useful goal or not, since they didn’t include mortality and morbidity information.

      • theadequatemother

        I would argue that these types of problems are endemic in health policy. This paper was reviewed, remember, and the editorial staff decided to accept it.

        It speaks to the philosophy of seeing knowledge and research as “strategy.”

        • auntbea

          Well, sure, but these problems are endemic in every field. Policy people do not have a monopoly on bad work.

          • theadequatemother

            I never said they did.

        • Mrs. W

          I’ll agree these problems are endemic and there is a tendency in health policy to see data for the strategic stories it can tell.

      • Laura

        “craptasticness” that is a great word auntbea! And a new one for me – love it

    • AllieFoyle

      It looks like they attempted to address the high risk vs. low risk issue, but still it’s mostly just data without the slightest attempt at meaningful analysis. Why did the rates differ? What other variables differed? What was the relationship with morbidity and mortality and patient satisfaction?

      • theadequatemother

        If I was to investigate this problem, this is my first stab at a research plan:
        1) investigate risk adjusted c/s rates between hospitals
        2) chose some representative cases and outliers and do a chart audit to verify the reliability of the billing data I used for step 1. If the data was unreliable I would chose a source other than billing and repeat step 1
        3) decide if there is a meaningful disparity or not
        4) if a meaninful disparity exists, I would do some case studies of high rate and low rate hospitals as well as some middle of the road hospitals looking for tentative explanations
        5) I would use data from 4 to refine my risk adjustment model
        6) I would validate my new risk adjustment model on a second sample
        7) I would see if a meaninful disparity exists.
        8) I would re-run the data on the second sample to see if there is any correlation between risk adjusted c/s rates and maternal and perinatal morbidity and mortality that is unexplained by the risk adjustment

        that’s at least 5 research papers and a couple of invited and noninvited review articles.

        • AllieFoyle

          But wouldn’t it just be easier to assume that c-section=bad and plot a bunch of points in an interesting shape and then suggest that we make policy changes?

          (I would love to see someone do all that and then pull it together for real analysis. I’d also like to see some study/discussion of how adding pressure to attain certain rates would affect outcomes, and *if* it were determined that lower rates would, on average, lead to better outcomes, how best to facilitate that without displacing some risk onto other people or restricting their rights)

  • Mrs. W

    Love this post! Somewhere along the way people have forgotten most of the point of maternity care and have replaced it with assumptions, and bad assumptions at that!

  • guest
  • Captain Obvious

    How long did it take them to do this study? Waste of everyone’s time.

    Slightly related, I cringe when homebirthers quote hospital CS rates on sites like I have seen people demonize a great hospital with a great NICU because their CS rate is above 40%, even though they do all the high risk deliveries and accept all the high risk transfers from nearby smaller hospitals. Meanwhile some little hospital with barely a nursery and no in house physicians that transfers out all their high risk patients to those previously stated high risk hospitals has a lower than 18% CS rate are the best place to deliver.

    • There was (or maybe, is) a rumor going around my county here in So Cal that our c-section rate is over 75%. I checked out the rates for all the hospitals, highest was around 36% – that particular hospital has a level III NICU and is also is a fave back-up for a lot of home birth midwives for transfers.

      • AlexisRT

        My favorite part is when they then claim the hospitals are lying in the stats. Uh huh. Although I also once had an acquaintance tell me her midwife told her that the official VBAC stats were wrong and they really did do them, which made no sense whatsoever. (The official stats had the whole hospital doing something like 3 a year and understandably she was considering changing hospitals.)

        I do take rates into account, but not simply in a “lower is better” way. When I chose a hospital for my 2nd (and wanted VBAC to be an option) I chose the hospital with the highest CS rate… but the 2nd highest VBAC rate. (The hospital with both the highest VBAC and lowest CS rate was ruled out because the OBs there practice according to Catholic health directives.) I happened to know that the hospital I chose received referrals from other hospitals.

      • Laura

        There was a physician (she might still be on staff) at a small community hospital near Orange County in Riverside County who was known to do c-sections at the drop of a hat. My sister was a nurse there for many years who had heard this. The hospital has a published cs rate that is through the roof. I highly suspect that that one obstetrician drove up those rates and that if you take her out of the data, it would be a more “normal” number.

    • Mrs. W

      My guess it didn’t take very long at all to do this study….

    • R T

      Right! Like when a group of women protested outside of the hospital where I spent months of my pregnancy in the Perinatal Special Care Unit. They were protesting the high csection rate! There’s a level III NICU, one of the biggest PSCU in the country and a helicopter pad on the roof because they bring in the most severe high risk pregnancies in from Hawaii and Arizona! My room was next to rooms of two women carrying quads! Of course there’s a high csection rate! We’re all just extremely lucky our babies and even ourselves LIVED! If I hadn’t been in danger of bleeding to death from my partial abruption, I would have walked out there and given them a piece of my mind!

  • Charlotte

    On topic: I loved my two unexpected c-sections and I’m looking forward to having another if I have a 3rd baby. I hate it when I see people using bad data to put them down, because the end result is always that c-section moms get tons of criticism for having one regardless of whether it was planned.

    Off topic: since y’all seemed to appreciate my last two Feminist Breeder updates and no one had complained about them yet, here’s another. She’s comparing the names of people who subscribe to her blog against her list of “trolls” and is turning people down and issuing refunds instead of letting them in. Is that even legal, especially since it’s not in the TOS that she can deny you based on your name alone?

    • I don’t have a creative name

      How in the world does she find any time to do her so called “attachment parenting”, with stuff like this? I would imagine combing through tons of information constantly just to keep away anyone capable of words other than “you rock, Mama!” would be simply exhausting.

      • Charlotte

        I wonder the same thing! She spends HOURS on the internet every day! There is no possible way she could be spending any meaningful amount of time with her children considering she spends literally all day every day policing her facebook. Even if she does it via phone app and is not at the keyboard, she is not present with her children if her attention is nearly always on her blog and page.

  • Kalacirya

    That’s some shoddy data analysis, that there. If I turned in plots like those, with their attached descriptions and nothing else, on an assignment, I’d get very little credit for the problem. Data analysis is not useful if by reading the analysis, I can’t get a general idea of what the data looks like in my head. If I had their description only, in no way would I imagine those particular plots.

  • In some disciplines, papers are retracted (or, even better, rejected) when the authors don’t understand statistics. Why is that not part of the peer review (or, even better, education) process in more of the medical literature?

    The figures actually make me think that the authors are almost intentionally deceptive. The distribution of data is obscured by spreading it out over some imaginary vertical line, and the “x-axis” certainly isn’t “number.” Heck, the data isn’t evenly distributed about some vertical midline! If you plot it like most people would plot data, a histogram with % deliveries on the x axis, you get something like the image below, which would make anyone sane say “oh, I wonder what the mean and standard deviation of that thing that looks like a normal distribution are.”

    (edit: note: I typed in the data by hand because the paper is behind a paywall and I’m at home)

    • auntbea

      Indeed. Those graphs are like what happens if you express a histogram a la a Rorshach blot.

      • desiree

        I don’t even want to say what I saw…

        • The Computer Ate My Nym

          Yes, but now that you haven’t said it, I saw it too.

    • Kalacirya

      A box and whisker plot would have been one appropriate way to represent this data, but I suppose those aren’t as pretty as the above plots.

      • I’m not so sure. My gripe isn’t with dot plots; it’s with the location of “# hospitals = 0”. Every dot plot I’ve ever seen (except for theirs, I suppose) has a more meaningful zero. Putting zero in the middle just doesn’t make numerical sense.

        • Kalacirya

          Well yeah, it’s a pretty dopey dot plot in my opinion. I’ve seen these before, but the plot in tandem with the misleading text is a bit too much for me.

          • auntbea

            I really think they just made the dot plot and then accidentally folded the paper while the paint was still wet.

    • DiomedesV

      “Why is that not part of the peer review (or, even better, education) process in more of the medical literature?”

      The inmates are guarding the asylum.

  • Lisa the Raptor

    Even me with my history degree and two college level math classes can understand how that is off like bad milk. Is it just me or are we getting the point where no one actually bothers to have things peer reviewed and double checked before the call in the media?

    • DirtyOldTown

      The cynical, dead part of my soul wonders if Project Hope has decided
      that there’s a lucrative donor pool to be mined among disaffected first
      world women who are unhappy with their birth experiences…

  • Love those two graphs. They really show where the big fat middle of the bell curve is and where the fringes lie.

  • fiftyfifty1

    Would love to have a more complete story here, maybe a qualitative study. Right now the media assumption is that differences in physician habits/culture are what is driving the different c-section rates. And that may indeed play in. But what else: maternal preferences, maternal risk factors, baby risk factors, transfer rates? Interesting that they define “lower risk women” as term, vertex, singleton and no prior c-section. That is still not a good comparison group. They need to break this group into first time mothers and mothers with a proven pelvis. A hospital where 50% of women are giving birth for the first time will (and should) have a higher c-section rate than a hospital with 75% of women giving birth to babies 2,3,4 etc.

    • auntbea

      What does a “qualitative study” mean in this context? In my field, it would mean a case study or other non-statistical analysis.

      • fiftyfifty1

        I guess I’m interested both in more quantitative info (e.g. primip rates, transfer rates etc) and qualitative info (interviews of physicians to find out opinions/attitudes, and mothers to find out about views on risk tolerance, values etc)

      • theadequatemother

        I would think that case studies of the outliers (in both directions) would be illuminating.

    • Elizabeth Abraham

      What else? Hospital staffing practices.

      The gold standard for l&d staffing is to have OBs work some days in clinic and some shifts in l&d, predictably. Also, dedicated on site anesthesia for the delivery ward. This way there’s always a physician there, and they can scrub in fast for a section, and there’s no rush to deliver if everything’s going well. If a mom has her baby in fifteen minutes, the doc finishes her shift. If a mom is still in labor at shift change, the fresh OB takes over.

      The problem is that this is very expensive. The hospitals that use this model tend to be large, tertiary care facilities, and most of their patients see a hospital based clinic for prenatal care.

      The alternative is for the nurses to page a woman’s ob in from out of hospital when she shows up in labor, or when the ob is needed. If you have to add travel time on to how long it takes to get to the or, wait and see can become very risky.

  • AlexisRT

    Some of the very high rates sometimes (not always, but sometimes) have clearly definable reasons. For example, the highest CS rate in my state? The Children’s Hospital of Philadelphia. The name should tip you off: it is not a standard delivery unit, but a special unit for mothers delivering babies with prenatally diagnosed complications. Understandably, the surgical delivery rate for these babies is high. (Although I am not sure what year this unit opened–it was only in the past few years–so it may not be included in this sample.)

    I don’t know if this happens so much in the US, but sometimes hospitals massage figures downward. A friend, when planning her birth in the UK several years ago, discovered that her local hospital (which no longer has full consultant maternity, but did then) had paper rates that looked wonderful… because they sent out all the potentially risky cases.

    • Bomb

      It is the same for Spokane, wa. All really high risk cases for most of eastern Washington and north Idaho go there. If you know your baby will have a serious heart defect, you go to Spokane. NCB people use the hospital stats to try to frighten women because they think half of mothers end up with sections, but it isn’t true at all.

      • Ceridwen

        I was thinking of Spokane when I read this post. I’m in North Idaho and if anything goes wrong during L&D for me or this baby I’ll be immediately shipped to Spokane. As would anyone for a pretty huge area around Spokane. This clearly impacts the stats, for both my local hospital (where the c-section rate is pretty low) and the hospitals in Spokane that wind up managing all the transfers. Yet very few people seem to understand this.

  • Sue

    The omission of outcomes data says a lot. By just comparing only crude rates, these people show that they have no idea about clinical policy – only ideological and financial. They have made no attempts to quantify cost-effectiveness – the discussion about quality is theoretical. WHere is the discussion about the compromise between neonatal outcomes and cesarean rates?

    So, we know what we already know – cesarean rates vary widely between hospitals. As do maternal demographics and risk factors.

  • “We shouldn’t be looking for an ideal average C-section rate. We should be looking for the C-section rate that produces the best outcomes. How does the perinatal mortality rate compare between hospitals with low C-section rates and high C-section rates?”

    This is exactly the right question.

    • Certified Hamster Midwife

      You know us women (other than Dr. Amy) – show us a bunch of numbers and our eyes glaze over.

      • LOL!

      • KarenJJ

        Speak for yourself. I won a prize in maths in primary school for my excellent accuracy in my times tables recitations.

        • Certified Hamster Midwife

          Whatever, harpy.

          • KarenJJ


        • Kalacirya

          But did you also win a President’s Award for fitness? Because if you didn’t, then I don’t really think I can take your comment seriously Karen.

    • I agree with you Alan, and this is precisely what I’ve been saying for years. I would add that studies like this also need to separate planned and emergency cesarean rates – and not just identify primary cesarean rates. Thanks to the 1985 WHO recommendation of 15%, plus an ideology by some that vaginal birth is always best unless it’s a matter of life or death, so much time, effort and money is wasted on looking exclusively at cesarean rates. We need to look at the wider picture – and at the outcomes that matter most to women and babies (i.e. mortality, morbidity and satisfaction).

      • I will totally credit information I have learned here in giving me a new perspective on this. It is not just weird crunchy people who are heavily biased toward vaginal birth, though, but also ACOG. My wife and I had to really insist on going for a scheduled second C-section rather than trying for a VBAC as the OB urged, and that seems a bit bizarre.

        • TiffanyEpiphany

          Trying to make new friends, huh?

          • I don’t have a creative name

            Hey, didn’t this blog help turn around some of your goofy ideas, too? If he’s thinking more clearly about these issues, great! I’m giving him the benefit of the doubt… for the moment. 😛

          • TiffanyEpiphany

            Yep. Sorry for stepping out too far with my comment.

          • You don’t say? What goofy ideas were those? 🙂

        • TiffanyEpiphany

          Alan, IDHACN pointed out to me that I was being perhaps a bit too hypocritical with my first comment to you here. Please excuse my poor attempt at sarcasm. This is a skill I just don’t have. Maybe I’m the one who has Asperger’s Syndrome.

          In any case, I’m sorry and I hope that you can, once again, forgive my impulsiveness.

          • No prob. Were you the one who downrated me, just out of curiosity? It doesn’t surprise me with many of my comments (not because *I* think they are bad but because I understand they will be controversial here); but on this one it seems odd. Maybe someone just plonks every comment I make on general principle? LOL

          • TiffanyEpiphany

            Yep, I downrated you here, and I’ve been downrating you elsewhere. It’s the only way I can hold my tongue with some of the stuff you say.

            But no worries. I’ll show more self-control from here on out.

          • KarenJJ

            The down rating is part of the new comment system. I’m not a huge fan of it.

        • Lisa from NY

          You should have listened to the OB. You obviously have no respect for doctors and their opinions.

          • Is this tongue in cheek? Very very dry if so.

      • carovee

        Did they even look at covariates? Were the high c-section hospitals teaching hospitals? Or, as others pointed out, hospitals that accept a lot of transfers?

    • ckdemommy

      I think that looking for a specific CS rate is flawed. We need to do research to determine what symptoms are best treated by CS and then allow doctors to make the judgment call on what would be the best course of treatment, not having to justify their choices to the hospital or insurance company. It;s up to the mother to decide whether she wants to follow that course of action or not, but the recommendation should be made based solely on what is best for the patient health-wise, not what will produce the “optimal” rate.