The connection between poor labour analgesia and PTSD

This is my second Mother’s day.  I don’t remember what we did last year.  I remember I had just gone back to work and was relieved to get out of the house and away from the baby to do adult things.  So maybe I wasn’t in much of the correct mood to celebrate Mother’s Day.  Mother’s day still feels like an occasion that applies to my mother, not to myself.

I don’t know about you (or your partners) but for me the process of becoming a mother was…well…unsettling is probably the best word I have to describe it.  The immediate postpartum period was a time when my most overwhelming feeling was one of abandonment and that no one was taking care of me at a time when I needed someone to because I had to take care of this little person who seriously needed something done for them every 1.5 to 2 hours day and night and I didn’t have time to look after myself.

I was completely unprepared for that.

It was a rocky transition, and I suspect that many women have a rocky transition to motherhood and I completely understand and sympathize with the desire to “do things differently next time” in the hopes of easing this transition.  “Doing things differently” might mean a different mode of delivery like attempting a VBAC or asking for a c-section or choosing a different provider or birthplace or eating differently or learning hypnobirthing or going to prenatal yoga or, my secret hope but totally impractical method – surrogacy.  I don’t know how effective “doing things differently” would be, but I do know that one of the toughest things to deal with is when the transition to motherhood is accompanied by mental health disorders such as depression, anxiety, feelings of trauma and post-traumatic stress disorder (PTSD).

When this study from Israel came out, it was promoted in the media as linking natural childbirth to a higher risk of PTSD (Science daily, American Friends of Tel Aviv, etc).  The study was then, predictably, trounced on Science and Sensibility by Dr. David White, a family practitioner in Toronto that focuses on maternity care, who felt that a) the study had dodgy methodology and b) that the study didn’t corroborate his clinical experience and was therefore dismissable.  He concludes that because he doesn’t find a similar proportion of his patients have symptoms or signs suggestive of PTSD, that this must not be a problem.  But I wonder, at your 6 week postpartum visit, are you looking?  Are you asking?  Or are you cooing over the baby and celebrating the birth and providing other social cues that would inhibit a woman from talking about feelings of trauma?  As a FP that does OB in low risk women that is friendly to intervention-free childbirth are you interacting with a subsample of women that might be lower risk anyway?

With respect to a (dodgy methodology) I completely agree.  The methodology was not robust enough to draw a firm conclusion that childbirth without pain medication increases the risk of PTSD.  The study is suggestive, nothing more.  With respect to b (no corroborating clinical experience) I take issue with this whole concept that if a study doesn’t confirm what I already think to be true, then it is probably not reflective of the truth.  Briefly, I am open to being surprised and discomfited by new information because that is how we learn.    It is much more powerful to look at this study within the context of other published studies on PTSD after childbirth.  So that is what we are going to do.  I’m going to go through the Polachek, Harari, Baum & Strous study in detail (because we all need practice in how to assess literature, it’s good for us, like vegetables and because I haven’t read this study until now) and then briefly review the PTSD after childbirth literature, the diagnostic criteria, discuss risk factors and hypothesized preventative approaches.  I want to be very clear here…this is a understudied area and we aren’t likely to find enough good evidence to make firm conclusions.  There is however, enough evidence to warrant further study in this area and a greater attention to this topic in daily clinical practice.

Polachek, Harari, Baum and Strous wrote a nice introduction that explains the rationale for their study.  No data on women in Israel, significant concerning data that PTSD is a problem after childbirth, not enough data on why it happens or what might increase the risk.  It is clear from this introduction that the study is exploratory.  The main questions seem to be, “do Israeli women suffer PTSD symptoms after childbirth at rates comparable to those reported in other areas of the world” and “let’s go looking for factors, described previously and one’s we just have a gut feeling about, that might influence the development of these symptoms.”

Inclusion Criteria:

Women were enrolled from those admitted to the maternity ward after a birth at the Tel Hashomer Medical Center as long as they were able to give consent.  The inclusion criteria are FUZZY and POORLY DEFINED.  Was it all women?  Was it women that came in spontaneous labour or were planned inductions (ie trials of labour) or did the sample also include planned (primary or repeat) c/s?  The study enrollment period was not defined.  The facility is a large tertiary care center.  The women’s hospital, which contains maternity, has upwards of 17 000 admitted patients a year.  This is a busy busy place.

The difficulty with fuzzy and poorly defined inclusion criteria is that it doesn’t allow us to understand the characteristics of the sample or whether the sample is biased.  For example, there is a birthing center in this facility.  Are those women included along with the ones in the higher risk labour and delivery unit?  Are some uncomplicated vaginal delivery patients allowed to go home after 6 hours of observation as occurs in some low risk units?  If so, those women wouldn’t be included in this sample.  How many women were approached but declined to participate?  This is all speculation, of course.  And without details we can speculate for hours about ways this sample may not be representative of the population.  If the study’s data is going to be used to make inferences to the population, the sample must be representative.  Maybe we could stop reading here because we can’t answer this important question about representativeness…or maybe it will be addressed because the authors have happily provided us with rich and detailed background and demographic information on their sample (sadly they don’t).

Instruments and Measurements:

At two days post birth for vaginal deliveries and 5 days post birth for cesarean deliveries, women who had agreed to take part were administered the following:

(note, I guess this might answer one of my questions above…a center that keeps women for two days after an uncomplicated vaginal delivery and 5 days after a c/s is probably not going to be happy letting someone go after 6 hours, or maybe it’s not culturally appropriate.  We keep our vaginal deliveries for 24 hours generally and our c/s for 3 days in Canada).

A “study inventory” consisting of psychosocial and demographic variables (not defined, not in an appendix)

A previously described, and presumably validated, relationship questionnaire

The international personality disorder examination (also previously described and presumably validated)

One month later, the women were contacted and administered the following:

A “study inventory” exploring responses and current mental state

the Post-Traumatic Stress Diagnostic Scale.  The PDS is a validated tool used to aid in the diagnosis of PTSD and to follow the severity of symptoms over time, treatment response etc.

Who wants more information?  I do I do! (waves hands, energetically).  What exactly was in those “study inventories?”  With respect to the one given at one month, it is implied, but not stated that it was crafted after looking at, and potentially analyzing the results from, the first set of questionnaires.   More information would be both helpful and reassuring to the reader here.

Statistical Methods

They used either the chi square test (to look at categorical data) or the two-sided t test (to compare means between two groups) but did multiple comparisons without adjusting their significance value from p = 0.05.  This leaves them open to making alpha error because the family wise error grows with each comparison you make.  If you are running 10 comparisons on the same set of data, you should really use something like the Bonferroni correction and set your p value to 0.05/10.  If you are running 10 comparisons on 10 different samples you don’t need to do this.  In this paper there were upwards of 40 statistical tests run.  The significance level should there be set somewhere around 0.05/40 or < 0.001.

Results

I’m just going to paraphrase here and I’m only going to include comparisons with a significance value of p < 0.001 as significant because otherwise I personally feel the risk of making a type I or alpha error is too high.  Of course, because this paper is fishing, further research could be done on any of the relationships they think they found and many of them might be important to the prevention, diagnosis and treatment of PTSD in new mothers.  102 women completed the first set of questionnaires and 89 completed the second set.  Only the 89 women that completed both sets were analyzed.

The math gets sort of confusing here.  At one month, 23 out of 89 or 25.9 % had “significant post-traumatic symptoms.”  This group included 3 women who met all the criteria for PTSD (3.4%) as well as 6/89 that were one or two symptoms short of a diagnosis or hadn’t had symptoms of long enough duration as well as 7 women that had significant symptoms without functional impairment and 7 women that had some symptoms and some functional impairment but where the severity did not reach the threshold for a diagnosis of PTSD.

For reference, the diagnostic criteria for PTSD can be read here and I recommend you do so because it is clear to me that even if most of these women didn’t meet the criteria, for 25.9% of women to be having some or many of these symptoms a month after giving birth is significant (clinically, not statistically)… although we need to take care because some of the symptoms on this list can also occur with sleep deprivation (irritability and other criterion D symptoms for example).  Symptoms from criterion B (intrusive recollection) and C (numbing/ avoidance) are certainly NOT part of the “normal adjusting to new motherhood experience.”  I would like to know how many of the 25.9% of women with post-traumatic symptoms had criterion B and C as well as or opposed to D.

For their comparisons they ended up with a group of 7 women who had full or partial PTSD (missing one or two criteria) vs 82 women who didn’t.  They created this group in order to have adequate statistical power and it had been done by researchers in this area in the past so there is precedent.  I don’t have a major problem with this.  I would have been better had they gotten more data so they could look at PTSD vs non-PTSD diagnosed women but sometimes in research you have to make compromises based on time and money.

So, weeding through all the comparisons, the only one with a significance level < 0.001 was the use of analgesia.  Women who had analgesia were less likely to fall into the PTSD/ Partial PTSD group compared to women that didn’t (p = 0.000).  Analgesia was not defined but I assume it means by any method: gas, narcotics by injection or IV or epidurals.  The authors don’t report the number of women in either the PTSD or comparison group that received analgesia.  They do have a confusing table that compares the PTSD group to the control group on the basis of mode of delivery that uses the term “natural” as an euphemism for vaginal where it is clear that for 2 women in the PTSD group they don’t know the delivery method because, once again, the math is confusing.  So, for their most significant finding they authors didn’t include the data in the paper.  I find this pretty disturbing.  I want to know the number of women in the PTSD and non-PTSD groups that used labour analgesia and what kinds were used.  Had I been a reviewer, I would have insisted on this.

Conclusions

I can support some, but not all of the conclusions the authors reached after analyzing their data.  I completely agree that 25.9% of women having these sorts of symptoms after childbirth is concerning and I don’t think we should sweep this finding under the rug.  We screen for PPD at discharge, at every public health nursing home visit and at the 6 week postpartum visit.  I personally feel that we should screen for post-traumatic stress symptoms too.  Social niceties are incredibly powerful and if we all assume that new moms should be over the moon happy and excited and enjoying every minute because new babies are so precious then they aren’t necessarily going to tell us if they are anxious or depressed or having nightmares or wanting to avoid their infant because it reminds them of the scary and painful experiences they had during the birth.  Think back on those tragic news stories where a new mother commits or attempts infanticide and/or suicide…how often are friends and relatives quoted saying they “just had no idea?”  Women experiencing depressive and anxious symptoms can hide them and we, as health care professionals, need to go actively looking for them.

The authors point out that those that treat and work with PTSD sufferers and PTSD advocates feel that the diagnostic criteria are too narrow and restrictive.  The incidence they found of PTSD after childbirth (3.4%) is in line with that previously reported.  There discussions regarding labour analgesia, respect for dignity and minimizing the undressed state, identifying those at risk (previous traumatic labours, high anxiety traits etc) and the potential use of more preparation (education, to minimize anxiety about the birth process) are important discussions that we should be having and while I think some of their conclusions are overstated and they have failed to intelligently discuss the methodological shortcomings of their study and deal with internal and external threats to validity, I really appreciate their willingness to engage with this topic.

TAM’s Brief Review of the Literature on PTSD in order to help put all of this in perspective:

Birth can be a traumatic event:

During painful births, women really do believe they are being irrevocably harmed (Boyce & Condon, Traumatic Childbirth and the role of deliberating. in Raphael & Wilson, Eds.  Psychological debriefing: theory, practice and evidence.  New York.  Cambridge University Press 2001.  Pg 272-80)

PTSD has been shown to develop after obstetric and gynaecologic care and procedures (Menage.  J Reprod Infant Psychol 1993; 11: 21-8)

It is the pain, not the injury that is perceived as traumatic (Schreiber.  Pain 1993; 54: 107-110) so it should not surprise us that pain without injury (like contractions) can be seen as traumatic.  Childbirth pain can be remembered as traumatic (Melzack.  Pain, 1993; 54: 117-120)

There are significant sequelae to traumatic childbirth:

Increased requests for future children to be born by c/s (Ryding.  Acta Obstet Gynaecol Scand 1993; 77:542-7)

Avoidance of sexual activity and fear of having future children, requests for sterilization (Fones.  J Nerv Ment Dis 1996; 184: 195-6)

Impaired attachment and breastfeeding (Ballard.  Br J Psychiatry 1995; 166:525-8)

Requests for termination of pregnancy (Godldbeck-Wood, BMJ 1996; 313: 774)

avoidance of other babies and mothers leading to increased maternal isolation and lack of peer support (Beck.  Nus Res 2004; 53: 216-224)

PTSD after childbirth is not uncommon:

PTSD symptoms are found in 6-24% of women and PTSD is diagnosed in 1-6% after childbirth (Czarnocka.  Br J Clin Psychol 2000; 39: 35-51)

PTSD occurred in 5.9% of Nigerian women at 6 weeks and was associated with emergency c/s, manual removal of the placenta, admission for pregnancy complications and loss of maternal control (Adewuya, BJOG, 2006; 113(3): 284-8)

PTSD occurred in 6% of this sample at 6 weeks, and nearly 15% by 6 months.  (Zaers J Psychosom Obstet Gynaecol. 2008 Mar;29(1):61-71.)

PTSD occurred in 8.6% of women in this study who had pre-ecclampsia (Hoedjes.  J Psychosom Obstet Gynaecol.  2011: 32(3): 126-34.

PTSD occurred in 7.6% of a sample of Canadian women at one month postpartum.  Partial PTSD occurred in 16.6% of women (Verreault.  J Psychosom Res 2012; 73(4): 257-63

A systemic review on anxiety disorders in the postpartum period: PTSD occurred in 0-6.9%.  25% of women had symptoms but didn’t meet the diagnostic criteria.  These women may be having what is called an “acute stress response” rather than a “disorder.”  (Ross J Clin Psychiatry 2006; 67(8): 1285-98)

Lots of things have been associated with increased risk of PTSD after childbirth:

invasive procedures

loss of control

fear of losing the baby

history of sexual abuse

previous psychiatric treatment for depressive or anxiety disorders

negative birth expectations

personality structure

previous traumatic life events (immigrants from war zones, traumatic experiences with family members)

forceps and vacuum deliveries > uncomplicated vaginal or c/s

Forceps or emergency c/s > uncomplicated vaginal or elective c/s

Pain

There are very few studies specifically looking at the treatment of PTSD and PTSD-like symptomatology in postpartum women.

There have been a couple of studies that show a positive effect for debriefing to prevent the development of PTSD in newly delivered women at risk (Ross, J Clin Psychiatry 2006 67(8): 1285-98)

For a disorder that affects between 1-6% of postpartum women, don’t you find it SHOCKING that there isn’t more research with respect to screening, prevention and treatment?

TAM’s Bottom Line and Childbirth-Associated PTSD Soapbox:

I wish more of those in the “birthing” world would engage with this topic. Instead of pretending that birth is all rainbows and unicorns and that will proper support and minimal interventions everything will go just fine, I wish we could have an intelligent adult conversation about the darker parts of the experience.  One quarter of women in the Israeli study experienced their birth as somewhat traumatic.  Sure, the sample could be biased, but if as little as 5-10% of women experience their birth as being traumatic it is definitely worth our attention.  And since around 3% of women experience PTSD after childbirth this topic ABSOLUTELY deserves more attention.  Let’s not forget, bringing a child into your family is supposed to be a happy time.  Women deserve delivery and postpartum experiences that make them feel supported and safe and taken care of.

Painful experiences are viewed as traumatic.  Childbirth can be very painful and the fact that someone hands you a baby at the end of it does not negate the trauma or perceived trauma of the experience.    Pain relief should be available on demand for women in labour.  And by on demand, I don’t mean in 4 hours when the only anesthesiologist in-house at night gets out of the OR.  Women should be counselled about their pain relief options and they should receive accurate information about epidural availability.  This is especially important in smaller hospitals, rural hospitals and areas where there is a shortage of anesthesiologists.  Women who think they want epidural pain relief in labour should be given the opportunity to choose to birth in larger centers with dedicated OB anesthesia coverage or higher staffing levels.  Women with previous traumatic births should be counselled that timely pain relief, may decrease their risk of developing post-traumatic symptoms with subsequent births.

When obstetrical emergencies happen, uncomfortable and painful procedures sometimes have to be done very quickly.  These include uterine massage, manual removal of retained placental fragments, vacuum and forceps deliveries, shoulder dystocia maneuvers etc.  Much as pain control eventually became an important part of initial trauma care in the emergency department, there should be a focus on this within LDR.  Labour and delivery units should audit these cases and develop plans and acquire resources for emergency pain control in these situations.

Women should be screened for post-traumatic symptoms in the postpartum period and formal community-based resources should be developed to help women experiencing distress.  Women should be screened for perceptions of trauma associated with their birth experience in the immediate postpartum period and be offered debriefing with their care team.

 

Check out the other posts in this second edition of the Carnival of Evidence-Based Parenting:

The Transition to New Motherhood (Momma, PhD)

Bonding in Early Motherhood:  When Angels Don’t Sing and the Earth Doesn’t Stand Still(Red Wine and Applesauce)

The Connection Between Poor Labour, Analgesia, and PTSD (The Adequate Mother)

For Love or Money:  What Makes Men Ready for New Fatherhood (Matt Shipman)

What the Science Says (and Doesn’t Say) About Breastfeeding Issues, Postpartum Adjustment, and Bonding (Fearless Formula Feeder)

No, Swaddling  Will Not Kill Your Baby (Melinda Wenner Moyer,  Slate)

Sleep Deprivation:  The Dark Side of Parenting (Science of Mom)

The Parenting Media and You (Momma Data)

Reassessing Happiness Research:  Are New Parents Really That Miserable? (Jessica Smock)

40 Long Days and Nights (Six Forty Nine)