The British Medical Journal (BMJ) has officially retracted an article that examined unexpected weight loss as a potential warning sign of cancer after the researchers identified an error in their work. Originally published on August 13, 2020, the study aimed to prioritize primary care patients experiencing unexplained weight loss for cancer investigations.
The original paper evaluated electronic health records to identify patients diagnosed with cancer within six months of being recorded with unexpected weight loss. However, while attempting to replicate their results in a different dataset, the authors found discrepancies that they could not easily explain. In their efforts to understand these differences, they discovered a significant selection bias caused by the exclusion of certain patients in the dataset.
Upon realizing the extent of this issue, the authors contacted BMJ to report their findings. They noted that some patients were inadvertently excluded from the study because their healthcare records did not initially indicate weight loss, even though they may have had a relevant diagnosis later on. This oversight led to an underestimation of cancer prevalence in the general primary care population.
On March 25, the BMJ published an expression of concern regarding the original paper. The notice highlighted how rectifying the selection bias would substantially alter the study’s findings and messaging, especially regarding the likelihood of cancer diagnosis among various patient demographics.
Ultimately, on October 16, the BMJ retracted the flawed article, replacing it with an updated version that presented revised findings. The new analysis included a larger patient dataset of approximately 326,000 individuals, which significantly altered the conclusions drawn about cancer risk following unexpected weight loss.
In the original study, only 1.4% of the nearly 64,000 patients with documented weight loss were diagnosed with cancer within half a year. The revised results indicate that 4.8% of the larger patient group received a cancer diagnosis, complicating the risk factors associated with unexpected weight loss. The re-analysis now advises referrals for cancer investigation for all men aged 50 and over, all women aged 60 and over, and other younger patients exhibiting additional symptoms.
The BMJ has taken proactive measures by notifying parties who cited the original work and updating relevant guidelines, ensuring that the new findings will inform future cancer referral practices in primary care settings.
Our hope is that these revised results will lead to more informed clinical practices and ultimately enhance patient outcomes in recognizing and addressing potential cancer indications.
Source: Retraction Watch
