Googling for a diagnosis—use of Google as a diagnostic aid: internet based study
BMJ 2006; 333 doi: https://doi.org/10.1136/bmj.39003.640567.AE (Published 30 November 2006) Cite this as: BMJ 2006;333:1143All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
I have myself used Google on a couple of occasions when presented with patients whose symptoms did not appear to present a diagnosis I was familiar with. One of those was a patient with purple urine in the catheter bag. After ruling out the standard potential causes of this such as consuming beetroot and reviewing the patient's medication chart I Googled 'purple urine'. The search produced information on Purple Urine Bag Syndrome (PUBS), a rare syndrome affecting chronically catheterised/constipated women who have alkaline urine or contract a UTI. Whilst it would not have affected the management of her UTI and subsequent change of colour of her urine to normal it was helpful to have a diagnosis to give to the patient. I doubt I would have found this through conventional book searches.
Whilst I feel Googling may be useful for rarer and eponymous syndromes doctors are less likely to come across in day-to-day practice, the diagnoses which can be offered are obscure and could worry patients or the doctors looking after them unnecessarily. It is a tool to be used with caution and with the same processes we use to formulate a diagnosis in regular practice.
The authors state that they selected the three 'most likely' diagnoses offered from Google. It would have been interesting to know what other diagnoses were suggested when they feel the search produced a match to know if these were similar or not, and if not have different the diagnoses were.
Whilst I appreciate this study was looking at the use of Googling for doctors, and thus they will have used their knowledge to select salient features to search with, and decide which 3 diagnoses were most likely, it would be interesting to give the same information to members of the general public and see how close their search terms were to those of qualified doctors, and also what diagnoses they would select.
Competing interests: None declared
Competing interests: No competing interests
Dear friends, We read with interest the article regarding Goggling for a diagnosis and agree with the conclusions that “Doctors and patients are becoming increasingly proficient with the Internet and frequently use Google to search for medical information” [1]. However, the article deals only with the diagnostic part of our work. As GPs we have the most challenging task in this rapidly changing information society, where we are presented with all types of questions regarding illness, well-being and health from a non selected population.
As GPs we have found great use of skills acquired as researchers which leads us to valid answers at the internet often during the consultation together with our patients. For the last years we have shared our experiences with our colleques in short intensive courses [2].
The systematic approach The obvious advantages of a systematic approach (the PICO model) when dealing with medical intervention lead us to categorize all types of questions we met in the clinic into four types: “Practice management”, “gate-keeping and cooperation within the health-care-system”, “diagnosis”, and “disease-management”.
Each question is further focused by describing on dimensions as; “time at your disposal”, “seriousness of the problem”, “local/national/global” and “relevant internet site/database”
How much time have you got? Don’t go on the internet, if you don’t have time to do so. Using the internet as inspiration is OK, if you have plenty of time. In our practice with a constant patient load, this is not possible. We have to “dance on the internet”, and skip the old virtue of going into details, and don’t get lost in the millions of hits at the web. Go straight for the answer, expect it within a minute, or else proceed and postpone seeking till later on. Save some time during the day and have a cup of coffee while finding an answer that satisfies your question.
Seriousness of the question? You have to interpretate the result and translate it into a decision. Is it an easy recommendation for common cold you are going to find? or is it a potential life threatening condition you should find an answer for. A Critical approach to the database source is today as important as the answer you find.
Is the wanted information of local, national or global nature? Is the information universal, will instructions from the neighbouring county suffice, as the question reflects national culture of treatment, or do you need to know, how to refer a patient to the medical ward?
Which databases do you use? The Google is good for some purposes, but not always sufficient. Pub Med, Cochrane, national databases and local instructions are also very handy and needy for the busy GP.
When the information is retrieved, the principles of the steps of EBM can be applied, and with information on the web becoming increasingly exhaustive and accessible to “lay” as to “pro”, the physicians ability to performing this will remain of crucial importance.
1. Tang H, Ng JHK. Googling for a diagnosis -use of Google as a diagnostic aid: internet based study. BMJ, doi:10.1136/bmj. 39003.640567.AE (published 10 November 2006).
2. Waldorff FB, Hermann C. [Rational decision making for the general practitioner: the effect of courses in Internet-based searches]Ugeskr Laeger. 2004 May 17;166(21):2054-7.
Competing interests: Conducting courses in internet searches for GPs
Competing interests: No competing interests
- ClusterMed
- Healia
- healthfinder: Consumer Health Information Search
- Healthline
- Kosmix
- Mamma Health Search
- MedStory
- PubFocus
- Scientific Commons
- Others: Search Engine Guide: Health Search Engines
Competing interests: Author/editor of the Medical Library Association Encyclopedic Guide to Searching and Finding Health Information on the Web
Competing interests: No competing interests
Madam,
Some correspondents have suggested that Tang and Ng’s Google searches (1) should be tried on PubMed.
I entered two or more ‘Google search terms’ from the study (column 2 in their additional table (2)) into the PubMed search window, successively in a Google-like fashion and in one line, until a set of (usually) less than some 50 references was retrieved. I then added the ‘final diagnosis’ to the search (column 4 in the table) and judged the PubMed search a success if two or more papers in the final set mentioned the correct diagnosis. Out of Tang’s 26 scenarios a PubMed search identified potentially relevant studies in 23 cases (88%), a better success rate than the corresponding Google searches (58%). I did not count references, if they occurred, to the relevant NEJM case studies.
I should mention that the searches on PubMed for this ad-hoc comparison were done without using MeSH terms, subheadings, or Boolean operators. No ‘nesting’ of concepts or use of double quotes was required. This is all done ‘in the background’ by PubMed’s very intelligent search and mapping algorithms, provided a succinct search formulation is entered. A search in the form
‘acute aortic regurgitation depression’ (case 5), ‘ulcerative colitis blurred vision’ (case 14), ‘sickle cell disease back pain’ (case 17), or ‘myopathy neoplasia dysphagia rash’ (case 29)
is perfectly acceptable and retrieves 8, 3, 21, and 4 studies respectively (out of 14 million+ records on PubMed). Other examples retrieve slightly larger sets, but they are always much smaller and more manageable than those retrieved in the corresponding Google searches. Indeed, in case 29 above, this crisp search formulation retrieves 4 papers, all of which mention ‘dermatomyositis’ – a ‘pathognomic question’ perhaps?
All 26 searches were replicated on PubMed in little more than one hour (about three minutes for each scenario).
Tang and Ng’s study may have some methodological weaknesses (it is for instance not quite clear how they selected the correct diagnosis from the Google search result); still their method of extracting specific discriminating concepts from a clinical synopsis is fascinating. They quote Wikipedia(!) and use the term ‘statistically improbable phrases’ (3), a principle used for the amazon.com search interface (4). It could perhaps also, somewhat archaically, be described as the ‘telegram style’ of searching (5): Take a clinical scenario and extract terms, which, if sent as a telegram, would be sufficient for the addressee to understand what your query is. One could also invoke Occam’s trusted razor (5,6); its ‘law of succinctness’ is still relevant.
Respectfully yours,
Reinhard Wentz
1 Tang H, Ng JH. Googling for a diagnosis--use of Google as a diagnostic aid: internet based study. BMJ 2006 Nov 10 http://www.bmj.com/cgi/rapidpdf/bmj.39003.640567.AEv1
2 http://www.bmj.com/cgi/data/bmj.39003.640567.AE/DC1/1
3 http://en.wikipedia.org/wiki/Statistically_Improbable_Phrases
4 http://www.amazon.com/gp/search-inside/sipshelp.html
5 Clarke J, Wentz R. Pragmatic approach is effective in evidence based health care. BMJ 2000;321:566-7 http://www.pubmedcentral.nih.gov/articlerender.fcgi?tool=pubmed&pubmedid=10968827
6 http://en.wikipedia.org/wiki/Occam's_razor
Competing interests: None declared
Competing interests: No competing interests
Competing interests: None declared
Competing interests: No competing interests
What fascinates me is how the mainstream media (in the US, anyway) reacted to this story. Headlines tended to read "Google diagnoses ..." when it is clear from the most superficial reading that the authors diagnosed and Google served as a clinical decision support; news stories often omitted that small detail about the 58% accuracy rate. Most interesting: complete omission of the fact that Google is not a content provider, has gone to court to stake its claim as a non-content provider, and so did not author the content that was used to make a diagnosis. However, the world speaks informally as if Google authored the content it provides.
It would have been much more useful an article if the researchers had given us information about the sources of the diagnostic answers--both the correct and the incorrect ones!
(Google "anti-SLAPP" and see Carafano v. Metrosplash.com, Inc., UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT, 339 F.3d 1119; 2003 U.S. App. LEXIS 16548) for caselaw on information systems and third-party content.)
Competing interests: None declared
Competing interests: No competing interests
I could not agree more with Dr Chamberlain about the usefulness of the internet in dermatology. I use it in my community dermatology clinic daily for decision support and patient education.
The most useful site by far is the New Zealand based www.dermnetnz.org. I get the pictures up and show the patient their rash and give them the url for further study if they wish to educate themselves more about their condition. This is particularly helpful in the all too common case of patients asking about the cause of their condition when it is unknown-I find they will take this better from an authoritative web site than from a humble GP.
I take the view that since my patients will use the web to obtain information about their condition, I should direct them to safe sites.
Google images is surpisingly helpful for dermatology and other conditions too, although safe search needs to be disabled which may be problematic if highly risk averse management fears that the practitioner may misuse the web and view unsuitable sites.
Long live medicine on the web, the only question is how can we make better use of it.
Competing interests: None declared
Competing interests: No competing interests
Google still does manage to surprise me at times, like a magician pulling things out of a magical hat, which is the internet. This is especially true when I am searching for information, which is not easily available. Despite the limitations of Google searching, it does make the task of searching for specific health information on the net relatively simple, and quite rewarding too. However, if you are looking for scientific health information not available on general medical websites, google scholar can be a good starting point, available at http://scholar.google.com/
Competing interests: Resource person for the National Informatics Centre, Delhi, India
Competing interests: No competing interests
After going through the article and the comments render by T C Winthrop and J. Martin Wehlou, we personally think if the electronic search engine like goggle is providing a platform that helps to diagnose various disease that require greater skill and experiences to diagnose, it should be well come by the medical community, as the success rate is 58% percent, mentioned in the article's authored by Hangwi Tang and Jennifer Hwee Kwoon NG. J. Martin Wehlou supported the article quite well as he is a MD and with the skills of an IT professional, but the comments of T C Winthrop should not be ruled out, as diagnosis life threatening disease should not be completely trusted on the data provided by any search engine. The search engines are now very much popular in developing countries and those are new to it. We think that the search engine and the professional skills should blend together for accurate and fast results in diagnosing various diseases.
Competing interests: None declared
Competing interests: No competing interests
Googling - Never the best to fall back on !
I am surprised and appalled to find that physicians and surgeons are resorting to internet-searches to arrive at a diagnosis. Although, the internet-based journals have been used hitherto for aiding in planning advanced therapeutic methodologies or diagnostic planning, the use in diagnosis of conditions is unethical. A doctor, be it a physician or a surgeon, is bound to be an astute clinician who knows his patient's history and condition better than anyone on the net. A diagnosis must therefore follow from the clinical acumen and experience of a doctor in the clinical setting and not from what happened in some other setting somewhere. Whatever be the accuracy of the document on the internet, it can never be equated to being the complete truth and since settings and patients vary considerably, it would be unethical to base your diagnosis on similarities found on the internet. Although, it would be easy to condemn such doctors, it is more prudent to check the over-dependence of medical students on the internet while at medical schools itself. Spare the rod and spoil the child !
Competing interests: None declared
Competing interests: No competing interests