Disclaimer #1: this is a Christmas edition of our regular journal club so expect it be as light-hearted as it is academic
Disclaimer #2: please do not proceed to read this blog post if you intend to watch the series, read the books or you have not watched all available episodes yet and you are an avid fan (spoiler alert!)
Disclaimer #3: as always you should read the original paper (it is open access so you have no real excuse) along with this blog before you draw your own conclusions
I have a confession to make: I do not watch much television…usually. This is a deliberate attempt to keep my free time available so I can do the things I believe are good for my wellbeing like practicing sports, socialising or travelling. It has become just so easy nowadays to come home from work and sit in front of the small screen or spend your free weekend watching garbage with no content in it. So I usually simply do not turn my television set on. It however happens now and then that I click on a new series on this or that channel and the inevitable happens: I become hooked and I struggle to leave the front of the screen.
The popular HBO television series Game of Thrones (GoT) is one of my current addictions and I have watched all seven series available…more than once!
For those of you who have not read the books by George R.R Martin or watched the series, GoT portray an imaginary island made up of seven kingdoms united in a single kingdom (Westeros) ruled by a single monarch. Like a United Kingdom of some sort. The series have everything one could ask for: political upheaval, civil wars, plots, dragons, love, allegiances, betrayals and a fair amount of sex scandals. This is a medieval society characterised by wide-spread violence so graphic portrayal of deaths is present in every single episode.
This is a rather lengthy introduction but I still have some close friends and colleagues who have not watched the series (fools!) so it was important that I set the scene before we have a look at the paper.
I have to confess I do not regularly read Injury Prevention (BMJ Journals) as it is slightly outside of my area of practice but it is where I came across the paper we will have a look at today. I also love the title. “Death is certain, the time is not”: mortality and survival in Game of Thrones.”
Remember that this a Christmas edition for our Journal Club so please bear with me and try keep a light approach on this rather than the usual academic one 🙂
What did the authors look at?
The primary aim of the study was to look at the mortality and survival rates of the main characters in GoT. They specifically wanted to estimate survival time and probability, identify predictive factors and describe causes and circumstances of death (I always wondered if the Dragon Queen would survive the full series and finally sit on the Iron Throne of Westeros so this was one of the main reasons I decided to dissect this paper)
The secondary aim was to give an excuse to the authors to watch the full series again (that is fair enough and anyone who disagrees with this will be thrown in the dark cells of the Red Keep)
The study included all important characters of the first seven seasons. Important characters were defined as anyone human (sadly the dragons are therefore not included), listed in the opening/closing credits (why don’t you listen to the amazing soundtracks in the background while you read this post?) , not already deceased when first appeared on screen (death is not that easy to establish in GoT but I will come back to this a bit later). Having a speaking role was not an essential requirement due to the high incidence of semi – elective and traumatic glossectomies (I recall at least one unfortunate character having his tongue ripped out in the throne room on the orders of King Joffrey). In summary, it is fair to say that the studied population was clearly defined by the authors pre-analysis.
The primary data source was the GoT DVD boxset (67 episodes in total) plus the Internet Movie Database. This could come under criticism as part of a critical appraisal as in the interest of internal validity the authors probably should have also reviewed the television version of the series and the published books. My colleague and friend Chris Gray tells me there is great discrepancy between the televised version and the books (I unfortunately got only to Book 2 so far so I am unable to further comment on this).
Sociodemographic variables were recorded and each character classified as highborn or lowborn in terms of social status, silk or boiled leather collar in terms of profession.
As mentioned above, the primary outcome was time to death. Death is usually relatively easy to determine in our profession so this should be easy to measure…one would have thought. Things however are not that easy in the Kingdom of Westeros as Jon Snow seems to have received great trauma care by the Red Priestess and survived several stab wounds to his torso after an established death. Survival time was based on the duration (measured in hours) that a character survived from when they were first introduced in the series.
For each death, the principal diagnosis, external cause of mortality and place of occurrence were recorded using the ICD-10-AM derived of the WHO’s classification list. For further details like physical location, time of day etc., please read the original paper.
This is probably the dry bit especially for a Christmas edition. Kaplan-Meier survival analysis was used to display survival experience of the studied cohort. Cox hazard regression was calculated to try and identify potential predictor variables. Final fitted multivariable model was constructed using time dependent covariates to identify independent predictors of survival and hazard ratios (HR) were calculated.
I bet you have been inpatient to get to this bit even if you are not well immersed in GoT!
320 characters were analysed (71.8% male, 68.5% lowborn, 10.6% followed the Faith of the Seven, 13.6% switched house allegiance during the series). For a full breakdown of the cohort characteristics, please refer to Table 1 in the original paper.
By the end of the study period (that is Season 7), 186 (56.4%) of the characters had deceased. If you had a favourite character who appeared in Season 1, chances are they did not survive the journey up to now! As expected, the majority of deaths were injuries (n=137, 73.7%) including 13 deliberate traumatic amputations at neck level (decapitation if you prefer). I still shiver at the public decapitation of Eddard Stark, Hand of the King.
The remainder of the deaths were chiefly burns (n=22, 11.8%, remember there are fire spitting dragons in the series) or poisoning (n=9, 4.8%). King Joffrey was poisoned on his own wedding (a great cinematographic scene you can re-watch here) and House Tyrell became extinct this way.
Only two deaths occurred from natural causes. I recall the old Maester of the Night Watch dying of old age in a warm bed but I am unsure of who the other was. Can anyone help with this so I do not have to watch the whole seven seasons again? 🙂
The most common circumstances of deaths were assaults (n=116, 63.0%), operations of war (n=45, 24.4%) making it a violent series in my opinion but don’t we all love a war scene with brave knights facing dragons?
A concise Table 2 breaks down the injuries into open wound to the head, intracranial injury, crushing injury to the head, open wound to the neck, fracture of the neck, open wounds to chest/abdomen, maltreatment syndrome, asphyxiation etc. Rather graphic. External causes are listed under assault, exposure to animate mechanical forces, legal interventions and operations of war etc. Rather descriptive. I have to admit that I have had quite a lot fun trying to fit the characters I remembered into one or two of these categories. Give it a go!
The survival time in hours for all important characters is displayed in a curve. This ranged from 11sec to 57h 17min with a median survival time of 28h 48min. The probability of surviving at least 1h in the show was 0.86 (95% CI 0.82 to 0.89). I would refer you to further curves contained in the article breaking down survival curves by sex, social status, whether the characters switched allegiances and how prominently they featured in the show.
Unsurprisingly, you were more likely to die if you were male (there was a majority of males in the studied population), lowborn (I speculate you were more likely to be sent to war or less likely to have a personal guard like the Queen Mother Cersei) or interestingly if you did not switch allegiance. This makes it for interesting premonition for any of your favourite characters in the show: will Arya Stark die then eventually?
The final fitted multivariable Cox proportional hazard regression model showed that the risk of death was more than 2.5 times greater if you were featuring prominently (HR 2.55, 95% CI 1.11 to 5.87, p=0.028). Characters who switched allegiance had significantly lower risk of death (HR 0.35, 95% CI 0.17 to 0.70, p=0.003).
The authors discuss that these results cannot be interpolated into human history and our current state of affairs somewhat limiting the external validity of the published results. The data suggesting that you were more likely to do die if you were male and lowborn however correlates with the real world suggesting that homicide rates are higher in countries and areas of lower socioeconomic status and 80% of homicide victims are male. They further discuss the absence of injury and violence prevention in the series but this is what makes GoT so attractive for viewers worldwide.
From a critical appraisal perspective this paper is still interesting as it helps us understand issues around gold standards, definitions, inclusion and exclusion criteria, statistics and more. Sometimes it’s easier to engage learners with an amusing study like this than something rather more dry and dull.
This is the first scientific study to examine the mortality and survival of the characters in GoT. This is also a great article for our Christmas edition as it is fun and entertaining to read regardless if you are fan or not.
The StEmlyns team would like to wish you a happy festive season and remember that season 8 is released in April 2019!
@baombejp on Twitter