Beausoleil or Bluepoint... or Phlegm?
Raw oysters on the half-shell. Just the words invoke strong responses. For some, these trigger dreamy-eyed memories of Beausoleil or Bluepoint delicacies garnished with favorite toppings. Others wrinkle their lips at the idea of eating a chilled, living, or recently dead filter feeder with the consistency of sturdy phlegm. Neutral feelings on eating raw oysters appear rare.
About fifteen years ago, I experienced a couple of the most challenging years of my life. One weekend, my father spent the day with me at one of our favorite places, the Getty Villa museum. Its architecture is a nearly exact recreation of Julius Caesar’s father-in-law’s home and perches on a cliff overlooking the ocean. This museum boasts the most extensive Greek, Roman, and Etruscan antiquities west of the Mississippi River.
Podcast: Free Audio File
If you prefer to listen to podcasts, feel free to play the audio version of this blog by clicking on the player above.
Podcast: The Three Little Pearls of Bias
Length: 6 minutes 53 seconds
Written and read by the author
How Do You Respond to Oysters?
My father and I leisurely strolled from room to room and chatted on the artwork and sculptures. In almost two years, I felt genuinely happy for the first time. Then my father decided to ruin this all by taking me out for oysters. You see, I was in the latter group of oyster eaters. With him so excited, though, how could I say no? We headed to Gladstones, a few minutes’ drive from the museum, and ordered a large, oyster sampling tray.
Since that day, my career trajectory shifted, my relationship with my father deepened, my financial situation healed, my health improved, and now, I adore oysters. Today, I find it hard to pass up any opportunity for oysters and am always reminded of my dad. Any meal with raw oysters gets “extra points” in my book.
We Judge Chefs on Their Ingredients, Researchers on Their Tests
As a result, I hold bias. Often, we consider bias in this context: love, hatred, dismissal. Wary of bias in studies, people will sometimes look at “disclosed conflicts of interest,” review who funded the study, and search for overly large p-values. For bias, these fields bear less fruit than you might think.
Bias arises with any systematic error that alters a value estimate through either design or analysis. Bias can work against the researcher and force acceptance of a null hypothesis just as much as it can overestimate a specific outcome. Despite a long list of specific types of bias, we broadly classify them into three buckets.
Chefs remain judged on the ingredients they use, and researchers, the tests. Information or measurement bias stands as our first type. Imagine comparing two populations who were screened with a test for a particular disease. The sensitivity for disease test kit A is 95%, while test kit B barely reaches 90%. To compare or integrate these results holds inherent flaws or bias.
Subpopulations Impart Bias
What if your study only used Test kit A, though? If you evaluate every individual with the same test sensitivity, can bias still result? Yes. Remember that bias occurs any time a systematic error skews our values away from the truth. Test kit A systematically misses five sick patients out of one hundred sick individuals tested. Ultimately, this bias may not generate enough impact to alter our perceptions, but bias nevertheless exists.
How studies pick their subjects imparts the second type of bias. Perhaps one breed, particular geography, or even a season may skew data away from reality. Any subpopulation naturally imparts bias. When mining hospital data regarding test kit A, did the test frequency dramatically shift with socioeconomic status? Proper subject selection is critical.
Poker-Playing Dogs Differ From Other Canines
The final type of bias is confounding. So confusing, science has literally named it “confounding.” To be a confounding variable it:
- 1) Holds a relationship with the risk exposure and the outcome, and
- 2) Is unevenly distributed within the test populations, and
- 3) Is not involved in the process of exposure generating disease
Perhaps a study gathered data to determine if dogs playing poker lived at a higher risk of cancer than those that do not. Initially, the data strongly supports the increased risk of pulmonary neoplasia for card-playing canines. The researchers, being astute, art lovers, noticed that poker-playing dogs smoke more tobacco than dogs who do not gamble. Incidentally, they tend to wear more bowler hats, too.
The biostatisticians stratify the smokers and voilà! The act of smoking increases cancer risk, but poker playing does not. From exposure to potential disease, confounding may divert our data and statistics away from the truth. Unless confounding is identified, repeating the study will still mislead. When identified and corrected, though, ante up, dear canines!
Like individuals having cuisine bias, most studies hold some type of scientific bias. Researchers work to limit bias as much as possible. As readers, we need to review each study for the three types of bias: information, selection, and confounding. We should view bias as a pearl of wisdom. Like an errant grain of sand in an oyster, its luminous image represents great value once found.
First Published in the Pulse Magazine
Dr. Christopher Lee pens the monthly column, Medical Leeway, in the popular Southern California VETERINARY MEDICAL ASSOCIATION (SCVMA) magazine, Pulse. The SCVMA published this article in their November 2020 magazine on page 15.
Access to these magazines is free and holds wonderful content. Whether you live in Southern California or not, consider joining the nation’s largest regional VMA. Veterinary and technician students can join for free. Become a member today!
References and Further Reading
1. Galarraga V, Boffetta P. Coffee Drinking and Risk of Lung Cancer-A Meta-Analysis. Cancer Epidemiol Biomarkers Prev. 2016 Jun;25(6):951-7. doi: 10.1158/1055-9965.EPI-15-0727. Epub 2016 Mar 28. PMID: 27021045.
2. Groenwold RH. 3 vormen van bias: vertekening van onderzoeksresultaten en hoe dat te voorkomen [Three types of bias: distortion of research results and how that can be prevented]. Ned Tijdschr Geneeskd. 2013;157(40):A6497.
3. Guertin, K. A., Freedman, N. D., Loftfield, E., Graubard, B. I., Caporaso, N. E., & Sinha, R. (2016). Coffee consumption and incidence of lung cancer in the NIH-AARP Diet and Health Study. International journal of epidemiology, 45(3), 929–939. https://doi.org/10.1093/ije/dyv104