We are awash in numbers, thanks in large part to the proliferation of personal mobile devices and the wrong-headed use of so-called big data.1 But applying statistical tools to the same set of data can support competing theories and lead to contradictory results. Such conflicting outcomes, known as antinomies if you remember Philosophy 101, cannot logically co-exist, and the field of statistics gets a bad reputation as a result. But big data can provide substantial value for people as individual patients. The key is to set some ground rules and understand the limitations of statistical investigation.
First and foremost, it’s important to gain some clarity regarding the concept of false positives in regards to health. This statistical construct is familiar to all of us, although we may not be aware of it. If one of your doctors sends you for a laboratory test and the results are “positive”, you’ll be sent for follow-up tests until a final determination is made. If the final test turns out “negative”, then the earlier results represented a false positive. The test results said you had the condition or disease, but in fact you did not.
False positives create numerous serious problems, not the least of which is the emotional toll of stress, anxiety, and fear experienced by the patient and her family and close friends. This is especially true when the suspected disease is a malignancy or other serious, life-threatening condition. It’s useful and empowering for people to learn that 5% of all test results are falsely positive right from the start. Medical tests are designed this way. The 5% false positive rate is a necessary part of statistical analysis. It’s built-in to the statistical design. In other words, test values that represent “normal” are obtained by cutting off the bottom 2.5% and the top 2.5% of a large sample of results from people who are “normal” for the thing being tested, such as white blood cell count or hemoglobin level.
Thus, 5% of normal people automatically have false positive results. Another way of stating this outcome is to consider that if you undergo a panel of 20 blood tests, one result (5% of 20) will be positive no matter what.
The vast majority of patients are not familiar with the statistical concept of false positive results.2 With a basic understanding of this construct and its implications, patients could ask their doctors meaningful questions such as, “What do the test results mean?”, “Have you considered the possibility of a false positive result?”, and “How will the additional tests you’re recommending affect decision-making in my case?”
Posing such questions is tremendously empowering for you, the patient, and helps reestablish equity in the doctor-patient relationship.3 As a health care consumer, a little knowledge goes a long way. Gaining more than a little knowledge by reading articles on diagnostic methods and health care decision-making will further strengthen your own process as a patient.
1Bates DW, et al: Big data in health care: using analytics to identify and manage high-risk and high-cost patients. Health Aff (Millwood) 33(7):1123-31, 2014
2Paddock SM: Statistical benchmarks for health care provider performance assessment: a comparison of standard approaches to a hierarchical Bayesian histogram-based method. Health Serv Res 49(3):1056-73, 2014
3Stacey D, et al: Decision aids for people facing health treatment or screening decisions. Cochane Database Syst Rev 28;1:CD001431, 2014