Against Initiatives: "don't be taken in by the boondoggle"

 ...or should I say braindoggle...


I've been reading The Future of the Brain, a collection of Essays by the World's Leading Neuroscientists edited by Gary Marcus and Jeremy Freeman. Amidst the chapters on jaw-dropping technical developments, Big Factory Science, and Grand Neuroscience Initiatives, one stood out for its contrarian stance (and personally reflective tone). Here's Professor Leah Krubitzer, who heads the Laboratory of Evolutionary Biology at University of California, Davis:

“From a personal rather than scientific standpoint, the final important thing I've learned is don't be taken in by the boondoggle, don't get caught up in technology, and be very suspicious of "initiatives." Science should be driven by questions that are generated by inquiry and in-depth analysis rather than top-down initiatives that dictate scientific directions. I have also learned to be suspicious of labels declaring this the "decade of" anything: The brain, The mind, Consciousness. There should be no time limit on discovery. Does anyone really believe we will solve these complex, nonlinear phenomena in ten years or even one hundred? Tightly bound temporal mandates can undermine the important, incremental, and seemingly small discoveries scientists make every day doing critical, basic, nonmandated research. These basic scientific discoveries have always been the foundation for clinical translation. By all means funding big questions and developing innovative techniques is worthwhile, but scientists and the science should dictate the process.”

...although it should be said that a bunch of scientists did at least contribute to the final direction taken by the BRAIN Initiative (Brain Research through Advancing Innovative Neurotechnologies^SM)...

from SLANG in the Great Depression

An AS @ UVA Project
by Meagan Hess
May 2004

Top image: vintage spoof Monopoly game issued during the 1936 US presidential campaign.

This Blog Is Brought to You by the Number 9 and the Letter K

The Neurocritic (the blog) began 9 years ago today.

I've enjoyed the journey immensely and look forward to the years to come, by Nodes of Ranvier (the band — not the myelin sheath gaps).

Node of Ranvier



And now a word from our sponsors,  Episode 3979 of Sesame Street...

The Number 9

The Letter k

Thank you for watching! (and reading).

Is it necessary to use brain imaging to understand teen girls' sexual decision making?

“It is feasible to recruit and retain a cohort of female participants to perform a functional magnetic resonance imaging [fMRI] task focused on making decisions about sex, on the basis of varying levels of hypothetical sexual risk, and to complete longitudinal prospective diaries following this task. Preliminary evidence suggests that risk level differentially impacts brain activity related to sexual decision making in these women [i.e., girls aged 14-15 yrs], which may be related to past and future sexual behaviors.”

-Hensel et al. (2015)

Can the brain activity of adolescents predict whether they are likely to make risky sexual decisions in the future?  I think this is the goal of a new pilot study by researchers at Indiana University and the Kinsey Institute (Hensel et al., 2015). While I have no reason to doubt the good intentions of the project, certain aspects of it make me uncomfortable.

But first, I have a confession to make. I'm not an expert in adolescent sexual health like first author Dr. Devon Hensel. Nor do I know much about pediatrics, adolescent medicine, health risk behaviors, sexually transmitted diseases, or the epidemiology of risk, like senior author Dr. J. Dennis Fortenberry (who has over 300 publications on these topics).  His papers include titles such as Time from first intercourse to first sexually transmitted infection diagnosis among adolescent women and Sexual learning, sexual experience, and healthy adolescent sex. Clearly, these are very important topics with serious personal and public health implications. But are fMRI studies of a potentially vulnerable population the best way to address these societal problems?

The study recruited 14 adolescent girls (mean age = 14.7 yrs) from health clinics in lower- to middle-income neighborhoods. Most of the participants (12 of the 14) were African-American, most did not drink or do drugs, and most had not yet engaged in sexual activity.  However, the clinics served areas with “high rates of early childbearing and sexually transmitted infection” so the implication is that these young women are at greater risk of poor outcomes than those who live in different neighborhoods.

Detailed sexual histories were obtained from the girls upon enrollment (see below). They also kept a diary of sexual thoughts and behaviors for 30 days.

Given the sensitive nature of the information revealed by minors, it's especially important to outline the informed consent procedures and the precautions taken to protect privacy. Yes, a parent or guardian gave their approval, and the girls completed informed consent documents that were approved by the local IRB. But I wanted to see more about this in the Methods. For example, did the parent or guardian have access to their daughters' answers and/or diaries, or was that private? This could have influenced the willingness of the girls to disclose potentially embarrassing behavior or “verboten” activities (prohibited by parental mores, church teachings, legal age of consent,¹ etc.). 

I don't know, maybe the standard procedures are obvious to those within the field of sexual health behavior, but they weren't to me.

Turning to more familiar territory, the experimental design for the neuroimaging study involved presentation of four different types of stimuli: (1) faces of adolescent males; (2) alcoholic beverages; (3) restaurant food; (4) household items (e.g., frying pan). My made-up examples of the stimuli are shown below.

Each picture was presented with information that indicated the item's risk level (“high” or “low”):

• Adolescent male faces: number of previous sexual partners and typical condom use (yes/no)
• Alcoholic beverages: number of alcohol units and whether there was a designated driver (yes/no)
• Food: calorie content and whether the restaurant serving the food had been cited in the past year for health code violations (yes/no)
• Household items: whether the object could be returned to the store (yes/no)

For each picture, participants rated how likely they were to: (1) have sex with the male, (2) drink the beverage, (3) eat the food, or (4) purchase the product (1 = very unlikely to 4 = very likely). There were 35 exemplars of each category, and each stimulus was presented in both “high” and “low” risk contexts. So oddly, the pizza was 100 calories and from a clean restaurant on one trial, compared to 1,000 calories and from a roach-infested dump on another trial.

The faces task was adapted from a study in adult women (Rupp et al., 2009) where the participants gave a mean likelihood rating of 2.45 for sex with low risk men vs. 1.41 for high risk men (significantly less likely for the latter). The teen girls showed the opposite result: 2.85 for low risk teen boys vs. 3.85 for high risk teen boys (significantly more likely) — the “bad boy” effect?

But the actual values were quite confusing. At one point the authors say they omitted the alcohol condition: “The present study focused on the legal behaviors (e.g., sexual behavior, buying item, and eating food) in which adolescents could participate.”

But in the Fig. 1 legend, they say the opposite (that the alcohol condition was included):

Panel (A) provides the average likelihood of young women's endorsing low- and high-risk decisions in the boy, alcohol, food, and household item (control) stimulus categories.

Then they say that the low-risk male faces were rated as the most unlikely (i.e., least preferred) of all stimuli.  But Fig. 1 itself shows that the low-risk food stimuli were rated as the most unlikely...

Regardless of the precise ratings, the young women were more drawn to all stimuli when they were in the high risk condition. The authors tried to make a case for more "risky" sexual choices among participants with higher levels of overt or covert sexual reporting, but the numbers were either impossibly low (for behavior) or thought-crimes only (for dreams/fantasy). So it's really hard to see how brain activity of any sort could be diagnostic of actual behavior at this point in their lives.

And the neuroimaging results were confusing as well. First, the less desirable low-risk stimuli elicited greater responses in cognitive and emotional control regions:

Neural activity in a cognitive-affective network, including prefrontal and anterior cingulate (ACC) regions, was significantly greater during low-risk decisions.

But then, we see that the more desirable high-risk sexual stimuli elicited greater responses in cognitive/emotional control regions:

Compared with other decisions, high-risk sexual decisions elicited greater activity in the anterior cingulate, and low-risk sexual decision elicited greater activity in regions of the visual cortex. 

This pattern went in the opposite direction from what was seen in adult women (Rupp et al., 2009), and it implicated a different region of the ACC. It's difficult to draw comparisons, though, because the adult and adolescent groups diverged in age, demographic characteristics, and sexual experience.

Figure adapted from Hensel et al., 2015 (left) and Rupp et al., 2009 (right).


So is it feasible to use fMRI to understand teen girls' sexual decision making? Maybe, from the point of view of logistics and subject compliance, which is no mean feat. But is it necessary, or even informative? Certainly not, in my view. It's not clear what neuroimaging will add to the picture, beyond the participants' fully disclosed sexual histories. Finally, is it ethical to use brain imaging to understand teen girls' sexual decision making? While the future predictive value of the fMRI data is uncertain, linking a biomarker to sensitive sexual information requires extra protection, especially when from a potentially vulnerable adolescent population.


Footnote

¹ In the state of Indiana, it is illegal for an individual 18 years of age or older to have sex with one of the participants in the present study. So if a young women engaged in sexual activity with an 18 year old senior, he could potentially go to jail. Not that this was necessarily the case for anyone here.


References

Hensel, D., Hummer, T., Acrurio, L., James, T., & Fortenberry, J. (2015). Feasibility of Functional Neuroimaging to Understand Adolescent Women's Sexual Decision Making. Journal of Adolescent Health. DOI: 10.1016/j.jadohealth.2014.11.004

Rupp, H., James, T., Ketterson, E., Sengelaub, D., Janssen, E., & Heiman, J. (2009). The role of the anterior cingulate cortex in women's sexual decision making. Neuroscience Letters, 449 (1), 42-47 DOI: 10.1016/j.neulet.2008.10.083

Interfering With Traumatic Memories of the Boston Marathon Bombings

The Boston Marathon bombings of April 15, 2013 killed three people and injured hundreds of others near the finish line of the iconic footrace. The oldest and most prominent marathon in the world, Boston attracts over 20,000 runners and 500,000 spectators. The terrorist act shocked and traumatized and unified the city.

What should the survivors do with their traumatic memories of the event? Many with disabling post-traumatic stress disorder (PTSD) receive therapy to lessen the impact of the trauma. Should they forget completely? Is it possible to selectively “alter” or “remove” a specific memory? Studies in rodents are investigating the use of pharmacological manipulations (Otis et al., 2014) and behavioral interventions (Monfils et al., 2009) to disrupt the reconsolidation of a conditioned fear memory. Translating these interventions into clinically effective treatments in humans is an ongoing challenge.

The process of reconsolidation may provide a window to altering unwanted memories. When an old memory is retrieved, it enters a transiently labile state, when it's susceptible to change before becoming consolidated and stored again (Nader & Hardt et al., 2009). There's some evidence that the autonomic response to a conditioned fear memory can be lessened by an “updating” procedure during the reconsolidation period (Schiller et al., 2010).¹ How this might apply to the recollection of personally experienced trauma memories is uncertain.


Remembering the Boston Bombings

Can you interfere with recall of a traumatic event by presenting competing information during the so-called reconsolidation window? A new study by Kredlow and Otto (2015) recruited 113 Boston University undergraduates who were in Boston on the day of the bombings. In the first testing session, participants wrote autobiographical essays recounting the details of their experience, prompted by specific questions. In principle, this procedure re-activated the traumatic memory, rendering it vulnerable to updating during the reconsolidation window (~6 hours).

The allotted time for the autobiographical essay was 4 min. After that, separate groups of subjects read either a neutral story, a negative story, or a positive story (for 5 min). The fourth group did not read a story. Presentation of a story that is not one's own would presumably “update” the personal memory of the bombings.

A second session occurred one week later. The participants were again asked to write an autobiographical essay for 4 min, under the same conditions as Session #1. They were also asked about their physical proximity to the bombings, whether they watched the marathon in person, feared for anyone's safety, and knew anyone who was injured or killed. Nineteen subjects were excluded for various reasons, leaving the final n=94.

One notable weakness is that we don't know anything about the mental health of these undergrads, except that they completed the 10 item Positive and Negative Affective Schedule (PANAS-SF) before each session. And they were “provided with mental health resources” after testing (presumably links to resources, since the study was conducted online).

In terms of proximity, 10% of the participants were within one block of the bombings (“Criterion A” stressor), placing them at risk for developing of PTSD. Most (95%) feared for someone's safety and 12% knew someone who was injured or killed (also considered Criterion A). But we don't know if anyone had a current or former PTSD diagnosis.

The authors predicted that reading the negative stories during the “autobiographical reconsolidation window” would yield the greatest reduction in episodic details recalled from Session #1 (S1) to Session #2 (S2), relative to the No-Story condition. This is because the negative story and the horrific memories are both negative in valence [although I'm not sure of what mechanism would account for this effect].²

Specifically, we hypothesized that learning a negative affective story during the reconsolidation window compared to no interference would interfere with the reconsolidation of memories of the Boston Marathon bombings. In addition, we expected the neutral and positive stories to result in some interference, but not as much as the negative story.

The essays were coded for the number of memory details recalled in S1 and S2 (by 3-5 raters³), and the main measure was the number of details recalled in S2 for each of the four conditions. Other factors taken into account were the number of words used in S1, and time between the Boston Marathon and the testing session (both of which influenced the number of details recalled).

The results are shown in Table 1 below. the authors reported comparisons between Negative Story vs. No Story (p<.05, d = 0.62), Neutral Story vs. No Story (p=.20, d = 0.39), and Positive Story vs. No Story (p=.83, d = 0.06). The effect sizes are “medium-ish” for both the Negative and Neutral comparisons, but only “significant” for Negative.

I would argue that the comparison between Negative Story vs. Neutral Story — which was not reported — is the only way to evaluate the valence aspect of the prediction, i.e. whether the reduction in details recalled was specific to reading a negative story vs. potentially any story. I wasn't exactly sure why they didn't do an ANOVA in the first place, either.


Nonetheless, Kredlow and Otto (2015) suggest that their study...

...represent[s] a step toward translating reconsolidation interference work to the clinic, as, to our knowledge, no published studies to date have examined nonpharmacological reconsolidation interference for clinically-relevant negative memories. Additional studies should examine reconsolidation interference paradigms, such as this one, in clinical populations.

If this work was indeed extended to clinical populations, I would suggest conducting the study under more controlled conditions (in the lab, not online), which would also allow close monitoring of any distress elicited by writing the autobiographical essay (essentially a symptom provocation design). As the authors acknowledge, it would be especially important to evaluate not only the declarative, detail-oriented aspects of the traumatic memories, but also any change in their emotional impact.


Further Reading

Brief review of memory reconsolidation

Media’s role in broadcasting acute stress following the Boston Marathon bombings

Autobiographical Memory for a Life-Threatening Airline Disaster

I Forget...


Footnotes

¹ But this effect hasn't replicated in other studies (e.g., Golkar et al., 2012).

² Here, the authors say:

...some degree of similarity between the original memory and interference task may be required to achieve interference effects. This is in line with research suggesting that external and internal context is an important factor in extinction learning and may also be relevant to reconsolidation. As such, activating the affective context in which a memory was originally consolidated may facilitate reconsolidation interference.

This is a very different strategy than the “updating of fear memories” approach, where a safety signal occurs before extinction. But conditioned fear (blue square paired with mild shock) is very different from episodic memories of a bombing scene.

³ Details of the coding system:

A group consensus coding system was used to code the memories. S1 and S2 memory descriptions for each participant were compared and coded for recall of memory details. One point was given for each detail from the S1 memory description that was recalled in the S2 memory description. Each memory pair was coded by between three to five raters until a consensus between three raters was reached. Raters were blind to participant randomization, but not to each other's ratings. Consensus was reached in 83% of memory pairs.

References

Kredlow MA, & Otto MW (2015). Interference with the reconsolidation of trauma-related memories in adults. Depression and anxiety, 32 (1), 32-7 PMID: 25585535

Monfils MH, Cowansage KK, Klann E, LeDoux JE. (2009). Extinction-reconsolidation boundaries: key to persistent attenuation of fear memories. Science 324:951-5.

Nader K, Hardt O. (2009). A single standard for memory: the case for reconsolidation. Nat Rev Neurosci. 10:224-34.

Otis JM, Werner CT, Mueller D. (2014). Noradrenergic Regulation of Fear and Drug-Associated Memory Reconsolidation. Neuropsychopharmacology. [Epub ahead of print]

Schiller D, Monfils MH, Raio CM, Johnson DC, Ledoux JE, & Phelps EA (2010). Preventing the return of fear in humans using reconsolidation update mechanisms. Nature 463: 49-53.

The Incredible Growing Brain!

The Incredible Grow Your Own Brain (Barron Bob)

Using super absorbent material from disposable diapers, MIT neuroengineers Ed Boyden, Fei Chen, and Paul Tillberg went well beyond the garden variety novelty store "Grow Brain" to expand real brain slices to nearly five times their normal size.

Boyden, E., Chen, F. & Tillberg, P. / MIT / Courtesy of NIH

A slice of a mouse brain (left) was expanded by nearly five-fold in each dimension by adding a water-soaking salt. The result — shown at smaller magnification (right) for comparison — has its anatomical structures are essentially unchanged. (Nature - E. Callaway)


As covered by Ewan Callaway in Nature:

Blown-up brains reveal nanoscale details

Material used in diaper absorbant can make brain tissue bigger and enable ordinary microscopes to resolve features down to 60 nanometres.

Microscopes make living cells and tissues appear bigger. But what if we could actually make the things bigger?

It might sound like the fantasy of a scientist who has read Alice’s Adventures in Wonderland too many times, but the concept is the basis for a new method that could enable biologists to image an entire brain in exquisite molecular detail using an ordinary microscope, and to resolve features that would normally be beyond the limits of optics.

The technique, called expansion microscopy, involves physically inflating biological tissues using a material more commonly found in baby nappies (diapers).

. . .

“What we’ve been trying to do is figure out if we can make everything bigger,” Boyden told the meeting at the NIH in Bethesda, Maryland. To manage this, his team used a chemical called acrylate that has two useful properties: it can form a dense mesh that holds proteins in place, and it swells in the presence of water.

Sodium polyacrylate (via Leonard Gelfand Center, CMU)

Acrylate, a type of salt also known as waterlock, is the substance that gives nappies their sponginess. When inflated, Boyden's tissues grow about 4.5 times in each dimension.

Just add water

Before swelling, the tissue is treated with a chemical cocktail that makes it transparent, and then with the fluorescent molecules that anchor specific proteins to the acrylate, which is then infused into tissue. Just as with nappies, adding water causes the acrylate polymer to swell. After stretching, the fluorescent-tagged molecules move further away from each other; proteins that were previously too close to distinguish with a visible-light microscope come into crisp focus. In his NIH presentation, Boyden suggested that the technique can resolve molecules that had been as close as 60nm before expansion.

Most scientists thought it was cool, but there were some naysayers: “This is certainly highly ingenious, but how much practical use it will be is less clear,” notes Guy Cox, a microscopy specialist at the University of Sydney, Australia.

Others saw nothing new with the latest brain-transforming gimmick. Below, Marc Schuster displays his 2011 invention, the inflatable brain.

“An inflatable brain makes a great prop for your Zombie Prom King costume,” says Schuster, author of The Grievers.


Link via Roger Highfield:

Gosh. That clever @eboyden3 has figured out how to blow up brains http://t.co/G8C17w3Uo0
— Roger Highfield (@RogerHighfield) January 9, 2015

The Futility of Progesterone for Traumatic Brain Injury (but hope for the future)

Traumatic Brain Injury (TBI) is a serious public health problem that affects about 1.5 million people per year in the US, with direct and indirect medical costs of over $50 billion. Rapid intervention to reduce the risk of death and disability is crucial. The diagnosis and treatment of TBI is an active area of preclinical and clinical research funded by NIH and other federal agencies.

But during the White House BRAIN Conference, a leading neurosurgeon painted a pessimistic picture of current treatments for acute TBI. In response to a question about clinical advances based on cellular neurobiology, Dr. Geoffry Manley noted that the field is on its 32nd or 33rd failed clinical trial. The termination of a very promising trial of progesterone for TBI had just been announced (the ProTECT III, Phase III Clinical Trial “based on 17 years of work with 200 positive papers in preclinical models”), although I couldn't find any notice at the time (Sept 30 2014).

Now, the results from ProTECT III have been published in the New England Journal of Medicine (Wright et al., 2014). 882 TBI patients from 49 trauma centers were enrolled in the study and randomized to receive progesterone, thought to be a neuroprotective agent, or placebo within 4 hours of major head injury. The severity of TBI fell in the moderate to severe range, as indicated by scores on the Glasgow Coma Scale (which rates the degree of impaired consciousness).

The primary outcome measure was the Extended Glasgow Outcome Scale (GOS-E) at six months post-injury. The trial was stopped at 882 patients (out of a planned 1140) because there was no way that progesterone would improve outcomes:

After the second interim analysis, the trial was stopped because of futility. For the primary hypothesis comparing progesterone with placebo, favorable outcomes occurred in 51.0% of patients assigned to progesterone and in 55.5% of those assigned to placebo. 

Analysis of subgroups by race, ethnicity, and injury severity showed no differences between them, but there was a suggestive (albeit non-significant) sex difference.

- click on image for a larger view -

Modified from Fig. 2 (Wright et al., 2014). Adjusted Relative Benefit in Predefined Subgroups. Note the red box p value for sex differences.


Squares to the left of the dotted line indicate that placebo performed better than progesterone in a given patient group, while values to the right favor progesterone. The error bars show confidence intervals, which indicate that nearly all groups overlap with 0 (representing zero benefit for progesterone) The red box indicates a near-significant difference between men and women, with women actually faring worse with progesterone than with placebo. You may quibble about conventional significance, but women on average deteriorated with treatment, while men were largely unaffected.

This was a highly disappointing outcome for a well-conducted study that built on promising results in smaller Phase II Clinical Trials (which were backed by a boatload of preclinical data). The authors reflect on this gloomy state of affairs:

The PROTECT III trial joins a growing list of negative or inconclusive trials in the arduous search for a treatment for TBI. To date, more than 30 clinical trials have investigated various compounds for the treatment of acute TBI, yet no treatment has succeeded at the confirmatory trial stage. Many reasons for the disappointing record of translating promising agents from the laboratory to the clinic have been postulated, including limited preclinical development work, poor drug penetration into the brain, delayed initiation of treatment, heterogeneity of injuries, variability in routine patient care across sites, and insensitive outcome measures.

If that isn't enough, a second failed trial of progesterone was published in the same issue of NEJM (Skolnick et al., 2014). This group reported on negative results from an even larger pharma-funded trial (SyNAPse, which is the tortured acronym for Study of a Neuroprotective Agent, Progesterone, in Severe Traumatic Brain Injury). The SyNAPse trial enrolled the projected number of 1180 patients across 21 countries, all with severe TBI. The percentage of patients with favorable outcomes at six months was 50.4% in the progesterone group and 50.5% in the placebo group.

The negative result of this study, combined with the results of the PROTECT III trial, should stimulate a rethinking of procedures for drug development and testing in TBI.

This led Dr. Lee H. Schwamm (2014) to expound on the flawed culture of research in an Editorial, invoking the feared god of false positive findings (Ioannidis, 2005) and his minions: small effect sizes, small n's, too few studies, flexibility of analysis, and bias. Schwamm pointed to problematic aspects of the Phase II Trials that preceded ProTECT III and SyNAPse, including modest effect sizes and better-than expected outcomes in the placebo group.


Hope for the Future

“And you have to give them hope.”
--Harvey Milk


When the going gets tough in research, who better to rally the troops than your local university press office? The day after Dr. Manley's presentation at the BRAIN conference on Sept. 30, the University of California San Francisco issued this optimistic news release:

$17M DoD Award Aims to Improve Clinical Trials for Traumatic Brain Injury

An unprecedented, public-private partnership funded by the Department of Defense (DoD) is being launched to drive the development of better-run clinical trials and may lead to the first successful treatments for traumatic brain injury, a condition affecting not only athletes and members of the military, but also millions among the general public, ranging from youngsters to elders.

Under the partnership, officially launched Oct. 1 with a $17 million, five-year award from the DoD, the research team, representing many universities, the Food and Drug Administration (FDA), companies and philanthropies, will examine data from thousands of patients in order to identify effective measures of brain injury and recovery, using biomarkers from blood, new imaging equipment and software, and other tools.
. . .

“TBI is really a multifaceted condition, not a single event,” said UCSF neurosurgeon Geoffrey T. Manley, MD, PhD, principal investigator for the new award... “TBI lags 40 to 50 years behind heart disease and cancer in terms of progress and understanding of the actual disease process and its potential aftermath. More than 30 clinical trials of potential TBI treatments have failed, and not a single drug has been approved.”

The TED (TBI Endpoints Development) Award is meant to accelerate research to improve TBI diagnostics, classification, and patient selection for clinical trials. Quite a reversal of fortune in one day.

Out of the ashes of two failed clinical trials, a phoenix arises. Hope for TBI patients and their families takes wing.


Further Reading (and viewing)

White House BRAIN Conference (blog post)

90 min video of the conference

Brief Storify (summary of the conference)

ClinicalTrials.gov listings for SyNAPSe and ProTECT III.


References

Schwamm, L. (2014). Progesterone for Traumatic Brain Injury — Resisting the Sirens' Song New England Journal of Medicine, 371 (26), 2522-2523 DOI: 10.1056/NEJMe1412951

Skolnick, B., Maas, A., Narayan, R., van der Hoop, R., MacAllister, T., Ward, J., Nelson, N., & Stocchetti, N. (2014). A Clinical Trial of Progesterone for Severe Traumatic Brain Injury New England Journal of Medicine, 371 (26), 2467-2476 DOI: 10.1056/NEJMoa1411090

Wright, D., Yeatts, S., Silbergleit, R., Palesch, Y., Hertzberg, V., Frankel, M., Goldstein, F., Caveney, A., Howlett-Smith, H., Bengelink, E., Manley, G., Merck, L., Janis, L., & Barsan, W. (2014). Very Early Administration of Progesterone for Acute Traumatic Brain Injury. New England Journal of Medicine, 371 (26), 2457-2466 DOI: 10.1056/NEJMoa1404304