The medical schemes at the centre of the “racial bias algorithm” furore didn’t conjure up fraud, waste and abuse cases out of thin air, writes Michael Avery in Business Day, reiterating the flawed logic in the arguments.
He writes:
Here’s a perfect reminder of why correlation isn’t causation. Sky Sports Cricket recently compared India’s Test match results with and without Jasprit Bumrah. The numbers imply that India wins more often when its star fast bowler is missing.
Does that mean Bumrah makes the team worse? Of course not. It’s a classic statistical illusion. Bumrah plays the hard overseas tours, while the “without Bumrah” record is padded with home victories.
The same flawed logic runs through the section 59 report on medical schemes’ fraud, waste and abuse (FWA) investigations. Just because black healthcare providers appear more frequently in fraud investigations doesn’t prove racial bias, and it has lit a fuse under the industry.
The report, tabled before Health Minister Aaron Motsoaledi two weeks ago, claims that in 2012-19, black healthcare providers were up to three times more likely to be flagged and sanctioned by FWA systems operated by Discovery Health, GEMS and Medscheme.
The panel calls this evidence of systemic racial discrimination.
Discovery and the Board of Healthcare Funders (BHF), representing 40 schemes and six administrators, have come out swinging, calling it flawed on several levels.
The panel’s statistical weapon of choice is the “risk ratio”, calculated by racialising provider data via surnames, a method Deloitte itself has previously described as “an inexpensive and efficient way” to estimate race, but hardly an exact science.
Add to this the inclusion of state hospitals and corporate entities, none of which is subject to the same FWA processes, and we have a data set that looks more like a thumb suck than a smoking gun.
Then there’s the elephant in the room: guilt. Schemes didn’t conjure up FWA cases out of thin air. Every case investigated proceeded because there was verified misconduct, often time-based coding anomalies that can’t be hand-waved away.
Claiming more billable hours in a day than exist in a day isn’t racial profiling; it’s arithmetic.
Discovery Health CEO Ron Whelan is correct to point out that no evidence was presented showing innocent providers being targeted because of race. The panel simply assumes that disparate outcomes must equal discrimination.
“They weren’t guilty because they were black,” as one insider quipped, “they were guilty because they were guilty.”
The panel seems to imply that prosecuting fraud among certain groups is inherently discriminatory, as if one’s background should entitle you to a free pass.
That’s a dangerously slippery slope.
The BHF’s counterpunch is worth quoting: first, exposure bias matters. A scheme with a membership base that is 91% black will, by definition, have more interactions with black providers. If more interactions occur, more investigations follow, regardless of race. The adjusted risk ratio drops to about 1.28 once this is factored in.
Second, the panel mistakes correlation for causation. Billing patterns, patient volumes and socioeconomic contexts all play a role in who gets audited. If you’re a Soweto-based physio charging scheme rates while seeing high patient volumes, you’ll naturally submit more claims and attract more scrutiny, just as a Sandton-based colleague would if they billed 200% of scheme rates.
Fraud, waste and abuse in the private healthcare sector costs about R30bn annually. And it’s reflected in your medical aid contributions, your benefits, and the steady upward creep of premiums. Discovery warns that neutering FWA systems under the guise of racial justice will only embolden fraudsters.
BHF MD Katlego Mothudi put it bluntly: “Corruption is corruption and must be rooted out.”
Yet the panel’s recommendations risk achieving the opposite. By framing data-driven fraud detection as structurally racist the report could make schemes hesitant to pursue cases aggressively, fearing accusations of bias.
The result? A bigger fraud bill for everyone. Scheme members, especially working-class families already stretched to breaking point, will end up subsidising those who game the system.
None of this is to deny that medical schemes have, at times, behaved like bullies. Their clawbacks are notoriously heavy-handed, and the acknowledgment-of-debt arrangements can leave smaller practices gasping for air. But procedural unfairness and racial discrimination are not the same thing. Fixing one doesn’t require inventing the other.
The section 59 panel’s reliance on statistics to draw sweeping conclusions smacks of confirmation bias. And frankly, who appointed Wim Trengove as a statistical oracle?
As any first-year stats student knows, correlation does not imply causation. There’s almost a 100% correlation between being guilty and being found guilty. Everything else is noise.
We can demand dignity and fairness for all healthcare providers without dismantling the mechanisms that protect scheme members from fraud.
Avery, a financial journalist and broadcaster, produces BDTV’s ‘Business Watch’.
Business Day article – Michael Avery: Healthcare probe: correlation is not guilt (Restricted access)
See more from MedicalBrief archives:
Top medical aid schemes guilty of racial bias, inquiry finds
Inquiry finds schemes discriminated against black healthcare providers