• OMSG Editor

Cracks in the Data: How Medical Research Contributes to Health Inequality

Updated: a day ago

By Hannah Pook, St John's College

We are taught medical research, and science as a whole, are objective pursuits. Scientists input questions and experiments output facts. Spending the third year of my medical degree ripping apart research papers has given me a different perspective. After knowing what to look for it isn’t hard to find cracks in the façade: this paper has used some dodgy statistics and that one didn’t include any controls. But something that really stuck out to me is the pervasive lack of representation. Pre-clinical studies commonly use only male mice, and genetic data is often overwhelmingly derived from Caucasian individuals. And the problem with these limitations is that they lead to more significant consequences than an unjust Nature publication. In fact, lack of representation has the potential to cause tremendous health inequality.


Historically, most clinical trials omitted women. This is now widely cited as a major ongoing cause of health disparities between sexes. Researchers trialling treatments considered women biologically distinct from men for reasons such as childbearing potential. But, bizarrely, they were then considered no different from men when new drugs or treatments were approved. One 1986 trial reflects the absurdity seen in many trials of that time. Researchers opted to examine links between obesity and uterine or breast cancer in a cohort of only men.


This bias extended into the pre-clinical stages of research as well. Scientists excluded female animal models because they introduced too much ‘variability’. Unfortunately, this practice continues to this day (Beery and Zucker, 2011). This widespread omission carries significant implications. Therapies better suited to women were more likely to be discarded, and those which evoked worse adverse drug reactions (ADRs) were more likely to be approved. Perhaps this is why until today women encounter ADRs nearly twice as often as men across all drug classes and have an increased risk of hospitalisation secondary to ADR (Zucker and Prendergast, 2020).


The 1990s marked a shift from this norm. The National Institute of Health (NIH) in the USA introduced the 1993 Revitalisation Act which required clinical trials to include women. This, alongside a later mandate to analyse results by sex, went some way in ensuring new drugs were both safer and more effective for women. However, one 2018 review casted doubt on their impact, showing that just 26% of 107 NIH funded randomised control trials reported any outcome by sex or included sex as a covariate in statistical analysis (Geller et al., 2018). Even over a quarter of a century after the NIH Revitalisation Act we still have a long way to go.

The Act also overlooked the continued use of therapies approved in older, flawed clinical trials. One such example is the sedative hypnotic drug zolpidem (brand name Ambien). For decades, clinicians and patients reported cognitive deficits as an adverse effect in women before the FDA introduced sex-based dose adjustments. Undoubtedly there are many other examples of how historical exclusion is still adversely affecting women.


Diversity in research extends beyond including both sexes. Ethnicity and race can also influence both pathology and treatment efficacy. Take the ongoing Covid-19 pandemic as a case in point. We now recognise that black and minority ethnic (BAME) individuals carry a higher risk of complications and death than white Europeans. Yet only two years ago just 13% of clinical trials reported outcomes by race or ethnicity according to one analysis (Geller et al., 2018). This could tell us why, despite having the highest rate of asthma in US, African American and Latino persons are less likely to benefit from inhalers than Caucasians. Or perhaps it could explain the much higher risk of severe, even fatal, adverse reactions seen in Asian individuals taking carbamazepine to prevent seizures (Devaney, 2019).


We are seeing similar failings regarding diversity in genetic research. In 2009, researchers at Duke University found 96% of genetic data was collected from people of European Ancestry. And as late as in 2016 White Europeans comprised 78% of those involved in genome wide association studies, a type of research examining pathological variants in our genomes.

Failing to diversify genetic research is especially likely to exacerbate health inequalities as the field of pharmacogenetics develops. This relatively novel area aims to personalise drug administration or dosages depending on an individual’s genome. Take the cancer therapy drug capecitabine as an example. We know it causes fatal toxicity in about 1% of patients. Genome studies linked variants in the DYPD gene to adverse effects, and dosage can now be personalised based on genetic data to make the treatment much safer (Deenen et al., 2011). However, many investigations, including that by Deenen et al, comprised mainly Caucasian patient populations. There are a huge number of DYDP variants, including one which was found only in African American individuals (Offer et al., 2013). So how can we be sure information about variants identified in Caucasian individuals applies to people of other ethnicities? Clearly we are far from gathering the level of genetic data we need to effectively guide treatment for non-Caucasian individuals.


Some researchers are making efforts to address the current lack of diversity in research. The NIH ‘All of Us’ Research Programme, for example, aims to gather health and genetic data from over a million diverse individuals. Currently, greater than 50% of participants belong to racial and ethnic minorities. Gathering this data has the potential to revolutionise healthcare for previously understudied populations, closing the health gap for many.


But collecting the data alone is just one step to bridging the gap. We need to analyse and interpret this data too. Disappointingly, new work published earlier this year indicated that even when diversity exists in genome databanks many researchers exclude information from ethnic minority individuals in their analyses, citing fears of confounding (Ben-Eghan et al., 2020). Out of 21 studies that used data from the UK biobank, 20 restricted their analysis to participants from majority groups. In doing so, they squandered the opportunity to gain useful insight into health and disease within minority populations.


Although things are changing, progress is happening a lot more slowly than we would like. Awareness of inequalities in medical research is just the first step. We, as current and future doctors, have a duty to our patients to make proactive changes now so that the way we carry out research does not continue to perpetuate current health inequalities for generations to come. Ideally, researchers would make a concerted effort to carry out studies aiming to abolish disparities between minority and majority populations. This could include, for example, specific research to produce more effective asthma drugs for those in African American and Latino groups. But even small changes, such as making sure clinical trial populations are reflective of the United Kingdom (or even the world) as a whole could make a huge difference to those currently impacted by lack of representation.



References

Beery, A. K. and Zucker, I. (2011) ‘Sex bias in neuroscience and biomedical research’, Neuroscience and Biobehavioral Reviews. NIH Public Access, pp. 565–572. doi: 10.1016/j.neubiorev.2010.07.002.

Ben-Eghan, C. et al. (2020) ‘Don’t ignore genetic data from minority populations’, Nature. Springer Science and Business Media LLC, 585(7824), pp. 184–186. doi: 10.1038/d41586-020-02547-3.

Deenen, M. J. et al. (2011) ‘Relationship between single nucleotide polymorphisms and haplotypes in DPYD and toxicity and efficacy of capecitabine in advanced colorectal cancer’, Clinical Cancer Research. American Association for Cancer Research, 17(10), pp. 3455–3468. doi: 10.1158/1078-0432.CCR-10-2209.

Devaney, S. (2019) ‘All of Us’, Nature. NLM (Medline), pp. S14–S17. doi: 10.1038/d41586-019-03717-8.

Geller, S. E. et al. (2018) ‘The more things change, the more they stay the same: A study to evaluate compliance with inclusion and assessment of women and minorities in randomized controlled trials’, Academic Medicine. Lippincott Williams and Wilkins, 93(4), pp. 630–635. doi: 10.1097/ACM.0000000000002027.

Offer, S. M. et al. (2013) ‘A DPYD variant (Y186C) in individuals of African ancestry is associated with reduced DPD enzyme activity’, Clinical Pharmacology and Therapeutics. Clin Pharmacol Ther, 94(1), pp. 158–166. doi: 10.1038/clpt.2013.69.

Zucker, I. and Prendergast, B. J. (2020) ‘Sex differences in pharmacokinetics predict adverse drug reactions in women’, Biology of sex differences. NLM (Medline), 11(1), p. 32. doi: 10.1186/s13293-020-00308-5.


Edition: 69 (2020-2010)

Correspondence to: editors@omsg-online.com

16 views

Recent Posts

See All

Get in touch!

Oxford Medical School Gazette

Medical School Office

Academic Centre, Level 2

John Radcliffe Hospital

Oxford, UK

© 2021 OMSG