What do the images of online doctors tell us about how doctors are portrayed in the Internet Age? What implications might this have on wider society and the medical profession itself? Hey Google! What’s my ethnicity?
This post supports my poster at the 2022 Midlands Racial Equality in Medicine Conference. Twitter @MREMnetwork
Learning objectives
1. Explore the image search results for different medical specialties
2. Compare online images with data on ethnicity
3. Reflect on what your observations might mean
I have previously explored the online appearance of doctors within internet image searches in prior posts, firstly in 2019 with ‘What does Dr Google look like?’ and then again a year later with ‘Google images, M.D. (class of 2020)’. In 2009, The BMJ also published my rapid response to an editorial titled ‘Ethnic minority staff and patients: a health service failure’ in the form of a letter ‘The ethnicity of doctors in photos online‘.
Questions
Why look at online images of doctors at all?
In a media saturated environment, images have increasing importance to the way information is shared. From cave paintings, stained glass, printing press, television, the internet and now social media, images have proliferated in volume and have never been so readily disseminated. These images not only shape how we understand the world around us but also how we perceive ourselves within that world. Insert an appropriate emoji here…
Google is the most popular internet search engine, processing over 90% of all search queries globally with over 40,000 search queries per second.

Although I have previously made comparisons with some other internet search engines, this post will focus on Google since it is far and away the most popular search tool online. Google Images searches was launched in 2001, apparently inspired by the high volume of regular Google searches for Jennifer Lopez’s green Versace dress worn to the Grammy awards in 2000. The former CEO and executive chairman of Google, Eric Scmidt said that “people wanted more than just text”.

The first page of Google reputedly captures 71-92% of search traffic clicks and the first five results are particularly important, receiving about 68% of clicks. The coveted first position has an average clickrate of 28.5% which falls considerably to 2.5% for tenth position.
As click-through rate (CTR) diminishes rapidly beyond the first few results, there is fierce competition for individuals, businesses and other organisations to appear high up the Google rankings in order to gain internet traffic and often revenue. Indeed, there is now an entire industry devoted to Search Engine Optimisation (SEO) that attempts to manipulate the algorithms Google uses to decide ranking positions.
Questions
How do you search for information?
Since my original post in 2019, the world has experienced a number of significant changes such as the Covid-19 pandemic which has a disproportionate impact doctors from certain ethnic groups. Cultural events have also occurred, for example, the Black Lives Matter movement as well as heightened awareness regarding gender inequality.
Let’s have a look at the fluidity in the top image results over the last 3 years for the term “doctor”:



The alteration in search results over time highlights that the Google algorithm, unlike a book, newspaper or television show, is not static. It is subject to modification, censorship and adaptive to user activity.
In this exploration, instead of looking at the more general term ‘doctor’, I instead performed searches for different medical specialties. I also wanted to see if there were any possible differences from real world data on the ethnicities of doctors working within those specialties.
Details regarding the NHS medical workforce were obtained via the ‘Hospital and Community Health Services (HCHS) Doctors by gender, ethnicity and speciality database’ (published 20/12/19, accessed 08/12/21). This includes data regarding doctors by gender, specialty and ethnicity, in NHS Trusts and CCGs in England between September 2009 to 2019. In order to keep things as contemporary as possible, I only used the most recent data from 2019 and discarded the rest. Data regarding General Practice is not included in this dataset and was therefore excluded from this investigation.
A Google Image search was then performed on the 07/01/22 for different specialty designations. The search terms were derived from the umbrella terms used by the HCHS database. For example, the HCHS designation ‘pathology group’ was changed to the search term ‘pathologist’. British spellings of specialisms were used, e.g. ‘anaesthetist’ instead of ‘anaesthesiologist’. The combined term ‘obstetrics and gynaecology’ was spilt into two different searches of ‘obstetrician’ and ‘gynaecologist’ as I felt that it would be more typical for individuals to search for these sub-specialisms individually than together. Indeed, Google Trends confirm this, with gynaecologist being the most searched term within this designation. Thanks Google.

Image searches were undertaken on the Apple Safari web browser in ‘private’ mode while using a virtual private network (VPN) to mitigate against the influence of personalised results. Screenshots for the first 10 images results for each specialty were taken.
Questions
How influential are personalised results when searching for images?
The image results are shown in the gallery below with HSCS data on ethnicity in the table underneath. Swipe through to see the different screenshot images while comparing it to the HSCS data.
HCHS data | Asian % | Black % | Chinese % | Mixed % | White % | Other % |
Anaesthetics | 23 | 2 | 2 | 3 | 59 | 11 |
Clinical oncology | 25 | 2 | 4 | 3 | 56 | 11 |
Emergency Medicine | 26 | 7 | 2 | 3 | 48 | 14 |
General medicine group | 27 | 4 | 3 | 3 | 50 | 14 |
Obstetrics & gynaecology | 29 | 8 | 2 | 3 | 50 | 14 |
Paediatric group | 27 | 6 | 2 | 3 | 49 | 13 |
Pathology group | 25 | 4 | 2 | 2 | 54 | 13 |
Psychiatry group | 30 | 6 | 1 | 3 | 48 | 12 |
Public Health Medicine | 23 | 3 | 1 | 3 | 59 | 11 |
Radiology group | 30 | 2 | 4 | 3 | 48 | 13 |
Surgical group | 29 | 4 | 3 | 3 | 46 | 15 |
Questions
Are there any differences between HCSC data and the Google Image results?
It would be tempting to project my own interpretations of the backgrounds of the individuals who appear in the Google Image results, or even try to quantify and compare them against HCSC data. However, the information regarding ethnicity from the HCSC database is based on how doctors self-identify whereas there is no such data about the individuals who appear on Google images.
Questions
What’s the difference between race and ethnicity?
The terms race and ethnicity are related but are often conflated. Here are two explanations:
Today, race refers to a group sharing outward physical characteristics and some commonalities of culture and history. Ethnicity refers to markers acquired from the group with which one shares cultural, traditional, and familial bonds.
The Difference between ‘Race’ and ‘Ethnicity’, Merriam-Webster
Race refers to perceived biological difference linked with physical characteristics such as skin colour and hair texture, whereas ethnicity refers to perceived cultural differences between groups.
Using the right words to address racial disparities in COVID-19
We are therefore comparing two different types of information: HCSC data on ethnicity and our own personal interpretation of the observed physical characteristics of individuals appearing on the image search results.
Designating ethnicity based purely on photographic appearance is challenging. Beside personal biases, technical issues such as viewer screen settings or indeed the physical properties of the ink used for posters may not accurately reproduce reality. Images are themselves vulnerable to alteration that may change the appearance of, for example, skin tone (e.g. filters). In the context of medical images, some specialities are prone to obscuring paraphernalia (e.g. masks, theatre lighting, microscopes, computer screens) making the health care worker’s appearance difficult to visualise.
For these reasons I have tried to avoid making detailed assumptions on the ethnicities of individuals who appear in the image search results. Instead, I want readers to reflect themselves, not only on potential disparities between the images and HCSC data but also the internal processes through which those interpretations are made.
The articulation of observations is also subject to variations in how different terminology is used and interpreted. For example, HCSC data includes self-identified ethnicities and the UK Government highlight that their preferred style is to refer to ethnicity and not race. This contrasts with the US, where there appears to be more emphasis on self-identifying using race but it is stressed that the racial categories used (in the census for example) generally reflect a social definition of race recognised [in the US] and not an attempt to define race biologically, anthropologically, or genetically. The use of collective terms (e.g. BAME or black, Asian, and minority ethnic) to describe minority groups are also insufficient as they negate the differences and histories within these groups.
There is no perfect terminology but it is up to us to discuss these issues and not allow a purely etymological debate to detract from the differences we might see between the reality of our everyday experiences and the pervasive pseudo-reality of the online environment.
Often people avoid discussing race/ethnicity and related issues due to fear of saying the wrong thing. Although a natural feeling, we must get better at talking about race even when it’s uncomfortable, to progress racial equality. Open conversations, including about language, are of great significance to both allyship and inclusion.
A guide to race and ethnicity terminology and language (The Law Society, 2020)
Again , I encourage viewers to make their own interpretations when comparing the Google image search results not only to HCSC data on ethnicity, but also their own real-life interactions with actual doctors. I feel it would be particularly insightful to reflect on the images of specialties where there are notably high numbers of doctors who identify as being from particular ethnic minority groups (e.g. obstetrics and gynaecology).
Questions
Do differences between real world data and medical imagery matter?
Beyond the concerns regarding the monopolisation of information, the ubiquitous use of Google as a means of mass information acquisition for the wider population may create a conflict between expectation and reality.
The terms epistemic bubble and echo chamber are also terms that are related but often conflated. The following definitions are taken from Nguyen:
- Epistemic bubble – A social epistemic structure in which other relevant voices have been left out, perhaps accidentally.
- Echo chamber – A social epistemic structure in which other relevant voices have been actively excluded and discredited.
It is unclear how the Google algorithm works and whether it creates a bubble, chamber or both. Although certain images may be available somewhere in the search results, we know that user interest and CTR fall rapidly beyond the first few items displayed. Lower ranked images are therefore essentially invisible for most realistic user searches.
A chamber of sorts is also created, although not through active exclusion but through intense SEO processes. Furthermore, Google itself promotes some results though it’s advertising programme called Google Ads. Financial power to promote certain results over others raises interesting thoughts into what images advertisers want potential customers to see and why. Google also censors some results (for example blurring images to address privacy concerns on Google Street View).
Whatever your views on censorship, blocking of sensitive queries, net neutrality or privacy may be, the core issue of this post is exploring the difference between reality and the internet when it comes to the depiction of doctors. I leave it to readers to reflect on the effect of internet images on patients, doctors and those seeking to enter the medical profession.
Limitations: Only the first 10 images yielded rom each search were included in the analysis and small numbers may skew findings, although data suggests that the first few results have the highest clickthrough rate and perhaps therefore are the most important and influential. The algorithms used by Google are complex and often updated to modify search results. Consider the changes that have already occurred since my initial exploration in 2019. Results may vary depending on the type of device used and geographic location. The algorithmic processes are also subject to manipulation by Search Engine Optimisation (SEO) processes as well as Google’s own advertisement programme. Variations of search terms may also yield different results (e.g. surgical doctor vs. surgeon).
Acknowledgements: I have no affiliation, partnership or endorsement from Google. I am grateful for the Google Brand Resource Center in clarifying the use of screenshot images for educational purposes: You don’t need to ask permission to use screenshots of the Google Search page and search results pages in print for educational or instructional purposes. Google and the Google logo are registered trademarks of Google LLC.
Quotes used in my virtual presentation supporting my poster at the 2022 Midlands Racial Equality in Medicine Conference:
Environments are not passive wrappings, but are rather, active processes which are invisible.
The Medium is the Massage (McLuhan & Fiore, 1967)
An image is synthetic. it is planned: created especially to serve a purpose, to make a certain kind of impression… it serves no purpose if people do not believe it.
The Image: A Guide to Pseudo-events in America (Boorstin, 1962)
We go to great lengths to make sure that imagery is useful, and reflects the world our users explore.
Google Maps
Media, by altering the environment, evoke in us unique ratios of sense perceptions. The extension of any one sense alters the way we think and act – the way we perceive the world.
The Medium is the Massage (McLuhan & Fiore, 1967)
The Spectacle is not a collection of images; it is a social relation between people that is mediated by images.
The Society of the Spectacle (Debord, 1967)
Google is dedicated to helping you discover the world around you. Imagery on our platforms is intended to enhance your experience, helping you preview and explore places nearby or across the globe.
Google Maps
Everywhere socialisation is measured by the exposure to media messages. Whoever is underexposed to the media is desocailised or virtually asocial.
Simulacra and Simulation (Baudrillard, 1981)
An image, like any other pseudo-event, becomes all the more interesting with our every effort to debunk it.
The Image: A Guide to Pseudo-events in America (Boorstin, 1962)
Historically, camera technology has excluded people of color, especially those with darker skin tones. A lack of diverse testing means that today’s cameras can carry that same bias, delivering unflattering photos for people of color.
Real Tone on Pixel 6, Google Store
The hyperreality of communication and of meaning. More real than the real, that is how the real is abolished.
Simulacra and Simulation (Baudrillard, 1981)
Google’s mission to make our camera and image products work more equitably for people of color. We vastly improved our camera tuning models and algorithms to more accurately highlight the nuances of diverse skin tones with Real Tone software. Pixel 6 Pro and Pixel 6 are the first phones with Real Tone.
Real Tone on Pixel 6, Google Store
In a world that is really upside down, the true is a moment of the false.
The Society of the Spectacle (Debord, 1967)
Summary
In our information age, misrepresentation of reality through the world’s most popular data gathering tool may alter not only the societal perception of the medical profession but also the profession’s understanding of itself. Depictions of doctors including differences in ethnic and/or racial background varies across specialisms and may show a disparity when compared to self-reported data on ethnicity. It may influence the decisions of doctors entering different specialities as well affecting doctors already working in these areas. The factors underlying and influencing internet search algorithms are complex and it raises questions about the potential biases involved for certain images to reach prestigious ranking positions.
Further resources
The ethnicity of doctors in photos online
https://www.bmj.com/content/365/bmj.l4370
Visual and textual research: Images of doctors
https://the-prescription.org/visual-research-doctor-images/
Using the right words to address racial disparities in COVID-19
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7373398/
The language of ethnicity
https://www.bmj.com/content/371/bmj.m4493