Especially when they don’t appear to be good research.
Want a clickbait headline to satisfy today’s tend for gendered outrage? Here’s one for you, straight from Gizmodo US: Women are more likely to die from a heart attack if their doctor is male.
A couple of choice quotes:
“The baseline rate of someone dying from their heart attack while in the hospital was 11.9 per cent. But women who were treated by men were 12 per cent more likely to die than the average patient, meaning their absolute chance of dying rose to 13.4 per cent. Among patients who survived, women treated by male doctors spent more time in the hospital before being released, further suggesting worse medical care.”
“Because the study is observational, it can’t provide any concrete answers as to why male doctors are flagging behind. But the authors do toss out some educated guesses. […] And, of course, given how the female doctors fared with all patients (along with research elsewhere), it could be the case that male doctors simply aren’t as skilled in general.”
“The study’s findings do suggest some tangible ways to improve care. […] Rather than rely on women to act as test dummies for inexperienced doctors, though, it’d better to just stock our emergency rooms and health care centres with more women doctors.”
Firstly, there’s quite a bit to unpack here – but before we do, can you imagine if the genders were reversed? How do you think the author would take it if an article showed that men fared worse when treated by female proctologists for prostate cancer, and suggested that it might be because they’re ‘less skilled’? This sort of gendered nonsense is pretty inflammatory, but before we crucify the article authors, let’s have a look at what the actual article says, and not the Gizmodo filtered version.
The original article
The original article (DOI: 10.1073/pnas.1800097115) was published in the Proceedings of the National Academy of Sciences of the USA – a multidisciplinary science journal. It isn’t specifically a medical journal. The journal article authors (Greenwood, Carnahan, and Huang) aren’t clinicians – they come from academic Business/Management backgrounds. So straight off the bat we can see that this isn’t a clinical study, it isn’t operated by clinicians, and it isn’t answering a clinical question. It is simply an observational ‘study’. That said, the fact that the authors aren’t clinicians and that it isn’t considered ‘high quality, practice-changing research’ doesn’t mean it isn’t without any merit. These sorts of articles are often the trigger for further research.
What was the aim of the study? The aim of the study was to observe if there was a relationship between clinical outcome in the setting of acute myocardial infarction (‘heart attack’) and the genders of the doctor and the patient. The introduction waffles on for a bit about general gender inequality across different dimensions of society, and cites an article that suggests female physicians perform better than males. Incidentally, the authors took a fairly interesting article and vastly overstated its significance – the study in question looks alarming initially, but if you drill into the article a few flaws emerge. There’s an editorial here that I don’t 100% agree with, but helps to shed some light on the flaws with the conclusions drawn. Importantly, while things appear statistically significant, the actual practical outcomes aren’t as significant – and ignore the fact that many of these patients are treated by multidisciplinary teams. But in any event, the authors seemed to set out to link outcomes from AMI with their doctor’s gender because… women face inequality? It seems like an attempt to focus heavily on the social aspects of medicine as opposed to the physiological aspects, but whatever.
How did they conduct it? The study looked at emergency department admissions between 1991 and 2010 in the state of Florida (US), using ICD 9 codes (a kind of code that identifies their diagnosed disease) – notably, the database they used included the name of their treating physician. The ED data was chosen because it creates a ‘discrete interaction’ between the patient and their treating doctor, and because patients can’t really choose their doctor at the ED. They examined men treating men, women treating women, men treating women, and women treating men.
The data they had access to included co-morbidities, age, race, gender, the name of their physician, when they were licensed, and clinical outcomes. The physician’s gender was inferred by their name, but ambiguous names were excluded (fair enough). Patients had to be in the ‘initial care’ phase of an AMI – that is, they couldn’t be continuing care, it had to be an acute onset. The primary outcome tested is simply ‘survival’ – that is they either lived or died. A secondary consideration was ‘length of stay’ which was basically how long it was before they were discharged from hospital.
The authors also tested the alleged impact of the environment on male physicians by assessing mortality rates based on how many female physicians were present in the practice environment at the time of treatment. They then suggest that this was appropriate because female colleagues “might also influence ER protocols in a way which helps the diagnosis and treatment of female patients” (Greenwood et al., 2017, p 4). They also looked at how many female AMIs the male physician had treated in the past – which I think is reasonable given female AMI symptoms are often ‘atypical’ compared to males.
What did they find? Statistically they note that having a female physician was associated with higher survival rates, both for males and females, and that male physicians that worked in areas with more female colleagues, or who treated more females for AMIs, seemed to improve their AMI patients’ survival rates.
What was the discussion about? The discussion noted the disparities between gender survival rates based on a male or female physician. It then states that the reason for this is “most physicians are male, and male physicians appear to have trouble treating female patients” (page 5). It also suggests that physicians still consider heart disease as a typically “male” condition – and cites an article from The Atlantic as proof (really?). To their credit, the authors note that while this analysis suggests a disparity, it doesn’t observe the cause of said disparity. Why does gender disconcordance (i.e.a difference between genders in the patient and their physician) result in different survival outcomes for females? Does the association with more female physicians or more female presentations suggest that male physicians become more familiar with atypical presentations? Maybe – the data doesn’t tell us. There was also this gem about omitted variables which might impact the results – “female physicians tend to perform better than male physicians across a wide variety of ailments.” Once again, the authors refer to the article I’ve linked above – which has its flaws. The authors rely heavily on this to suggest that the “most skilled” physicians are therefore female (page 5). Finally, the authors admit that the ‘treating physician’ is actually the ‘supervising physician’ and they couldn’t observe the effects of the supporting physicians (e.g. residents, nurses, other physicians providing input like cardiologists or cardiothoracic teams), and thus can’t comment on their input.
Is it a good article?
Well, it’s pretty meh, to be honest. To be fair, it’s not a randomised control trial or a systematic review, so it’s not going to change practice and has no real significance to the practice environment. In other words, nobody’s going to start assigning males to males and females to females in the hospitals. So I don’t think it’s fair to criticise it too much from a clinical standpoint.
The vast majority of the opening segment, and the apparent reason for the research, is couched in social justice terms. That’s fine, and it’s okay to point out inequalities – but the significant issue I have is that the authors rely heavily on this perspective to examine something which also has an underlying clinical issue as well. None of the authors appear to have any clinical background. In fact, if you check out their profiles, they’re very much rooted in ‘entrepreneurship’ and social justice movements. As such this article isn’t much use for clinicians, and attempting to link it to clinical practice isn’t going to stand up to scrutiny. I’m not accusing them of bias – at least not in the study’s intent.
The data is interesting, but again, we’re only seeing a correlation at best – and not causation. We don’t actually know the reasons behind this statistical relationship – something that people who love a clickbait headline will quickly forget. I have a few issues with the study.
Firstly, a binary ‘did they live?’ outcome isn’t a great metric to use. For example, a recent double-blinded randomised controlled trial called PARAMEDIC-2 looked at adrenaline (epinephrine) given in the prehospital environment for cardiac arrest. By the statistics, those who got adrenaline instead of a placebo had a higher return of spontaneous circulation rate. But by clinical outcome of neurological deficit, those who got the placebo had better neuro outcomes than those who got the placebo. Absolute survival can mean a lot of things and doesn’t really describe function after the fact. The time to discharge (i.e. Length of Stay) is probably a better metric because it may suggest the clinical progression and recovery. That said, we don’t know a lot about what sort of infarcts these patients had – the degree of coronary occlusion, the expected clinical outcome, the treatment modality, etc. While elements such as age and a list of co-morbidities were included as modifying factors, lumping them in together creates a muddled clinical picture, and makes it harder to draw clear distinctions on clinical outcomes.
Basically, this is just an interesting data point – because outside of a broad generalisation, we can’t really draw any conclusions at all from the data.
There are a few other issues. Firstly, it takes a team to save a life – no patient in the ED is ever treated by only a single physician, especially for an acute MI. For example, you’ll be assessed by nurses, several doctors, probably have cardiology input, and may end up going down several clinical pathways – with loads of other clinicians on the way. Maybe you get fibrinolysis right then and there, on the bed. Maybe you end up getting taken up to cath lab with a stent put in. Maybe you died because the cardiologist messed up – maybe they were female too. Maybe the triage RN didn’t triage you appropriately, and maybe they were female. There’s a massive list of factors other than ‘supervising physician’s gender’ that could be contributing to the disparity that we don’t know about.
Finally, and my biggest issue, is that the authors rely on Tsugawa et al. (2017) for the suggestion that female physicians are better than males at a ‘wide range’ of presentations. As previously noted the practical applications of that article are somewhat limited and don’t actually suggest such a sweeping generalisation – so I’d argue that the authors were too quick to accept it as fact.
I find their assertions to be pretty variable – and mostly unsupported by quality evidence, save for the fact that females are more likely to present atypically. I’ve seen this in my own practice as a paramedic – I’ve attended an 80 year old female having a massive inferior STEMI who was only complaining of vomiting. She waited 2 days to contact anybody because she assumed it was just food poisoning. This is a common tale – across both men and women – in my environment. The assertion that men are worse doctors than women because of this data and one article they’ve cited isn’t borne out by the data – and shouldn’t be a factor for healthcare consumers (i.e. patients) when they roll into hospital. There are too many other factors to make such a clear distinction.
That said, I don’t think it’s an awful article by any stretch – it’s fairly well written, reasonably well designed, and asks an interesting (if somewhat loaded) question. The pseudo-randomisation aspect might be debatable. Beyond that, I don’t know if it has much merit.
Gizmodo’s Dumpster Fire Article
I absolutely hate it when Gizmodo reports on medical research – because the authors invariably go for clickbait headlines and tiny tidbits without apparently reading the article properly (or relying on somebody’s summary – often somebody who isn’t a clinician). The Gizmodo article includes a few quotes from the authors.
Carnahan makes one assertion: that “about 1500-3000 fewer of the female heart attack patients would have passed away” if they were treated by female doctors. Perhaps – but again, we don’t really know too much about these patients from the data set. That’s an assertion made purely from the survival rate and is a common trap found with medical research – ‘the number says x therefore if we’d done y, x would have survived.’ Maybe – that depends on how specific the question is, the strength of the study design, and the population assessed. This was a fairly generalised study that ultimately tried to use a load of factors to focus on gender discordance/concordance and survival. It makes for a nice sound byte to list big numbers like that, but that’s a massive range, and the suggestion that they ‘would not have passed away’ might not be borne out in actual practice. This is really a meaningless statistics – similar to how people have reported on the PARAMEDIC-2 trial, with misleading headlines like ‘adrenaline hurts patients’.
The Gizmodo commentary is complete garbage.
- The leading paragraph states the article ‘suggests that men, who account for the majority of doctors, are worse at treating female heart attack patients.’ A nice attention grabbing headline, especially if you want your daily dose of outrage and gender politics. Except the data doesn’t suggest that at all – just that survival rates are lower, for any number of reasons.
- It asserts that ‘men might be worse at communicating with females.’ Except the study doesn’t suggest that and nothing in the data suggests that – because communication wasn’t examined.
- It suggests that male doctors ‘simply aren’t as skilled in general’ based on this data (and ‘research elsewhere’, which is almost certainly the same article we’ve already discussed). Except the data doesn’t actually suggest that, because competence isn’t measured by simple survival and we don’t really know the factors that led to a disparity in survival.
- To be fair, the article does note that the gender difference ‘probably’ doesn’t explain the gap between heart disease survival rates. Except that the data doesn’t explain it at all – the data explains nothing, it’s just a simple observation with any number of potential explanations.
- The article goes on to quote some of the suggestions for improving outcomes, along with a suggestion that ‘it’d be better to just stock our emergency rooms […] with more women doctors’. Except that’s a crap attitude towards education (and counter to how clinical education works), and while the data suggests a casual relationship between experience with female AMI presentations or more female colleagues being present, it doesn’t suggest it as a causative factor.
The Gizmodo article was written by Ed Cara, who writes about a lot of science articles. His LinkedIn profile lists ‘Civil Rights and Social Action’ as causes he cares about – which is probably why his article so strongly pushes the ‘male physicians are bad’ biased viewpoint, when the actual authors, while slightly tending in that direction, at least noted that such a conclusion was impossible to draw from such an observational study. I can’t find if Ed Cara has any actual clinical qualifications or credentials, so I can’t say whether or not he properly reads all of the articles he comments on, and whether that commentary has any weight. Maybe you could say the same about my commentary here – I’ll take that criticism. I’m not a medical doctor, after all – but I don’t think Cara is either, and neither were the authors of the original journal article.
The clickbait headline, the carefully selected, eye-grabbing points, and the overall tone clearly paint the picture that the author was hoping to demonstrate: that men are bad doctors. Unfortunately, reading the actual article demonstrates that the observational study doesn’t prove this at all – it just demonstrates a casual relationship, not a causative one, of survival and gender discordance between patient and physician. The reasons for it aren’t clear, aren’t demonstrated by the data, and aren’t even worth commenting on – any assertions beyond that are pure guesses, mostly relying on a single article which has its own issues.
But the Gizmodo article, while it occasionally makes concessions (“[…]it can’t provide any concrete answers as to why male doctors are flagging behind”) it’s clear that Cara was out to push the agenda that male physicians are bad compared to females. The authors themselves don’t go to those lengths – they make mere suggestions but still acknowledge that the data doesn’t really demonstrate that at all. I guess that’s not good enough for Cara and Gizmodo. I find it hilarious that Cara would even entertain the idea that educating ‘inexperienced doctors’ was a bad idea compared to just hiring more women – even though it isn’t clear why female physicians were linked with higher survival. The very idea that education is inferior compared to just hiring someone based entirely on gender, as if it’s the only factor that affected anything, is fucking ridiculous. Granted, the authors did make a similar assertion (with dubious evidence) in their own article, but Cara ran with it.
In short, it reads like every other social justice-laden Gizmodo article of late – overblown, biased, pushing an agenda, and not interested in the actual story. And this shit gets imported on the Gizmodo AU site.
It’s an awfully written article that should have been flagged as an opinion piece instead of educated commentary. Do better, Gizmodo.