Recall that the lumbar surgery posting was about a freshly minted JAMA article that found that while the overall rate of back surgeries among Medicare beneficiaries was unchanged, the relative proportion of complex and expensive surgery had increased. The authors concluded that this was an example of profits being put before patients. The DMCB was less sure about that and pointed out that there may be some sources of research bias.
c3 also looked at the article and pointed out.......
The confidence interval, EXCEPT FOR COST (and rehospitalization in 30 days), for most complications generally overlapped (Readers can look at the data table here)
All of the surgeries (CPT codes) are covered by Medicare. And since Medicare doesn't do prior auth, its all a moot point.
Even though the confidence intervals for overlapped, the authors based their conclusions on p values. This is outside the DMCB's statistical expertise, but suspects this has something to do with non-normal distributions involving odds ratios. That being said, the conclusion that the complication rates for more complex surgery are lower may be not quite correct.
As for the other DMCB posting on what docs can do to improve the health care system, c3 turns to the same seven point system used in the article and provides additional commentary:
1) Work Daily to Provide High Quality Care
But be prepared to prove it
2) Control Costs
At its core no business is in the business of reducing revenue. As much as I'd like docs to always be pre-disposed to save the system $$ I know they're not inclined to do so. However, if they provided pricing transparency and the patients had a financial stake in the game, that would make difference.
3) Improve Communication
Amen. But EHR's need to talk to each other. And we still haven't mastered the management of so much more info
4) Become Involved Locally
A nice sentiment but it feels pretty mom and apple pie
5) Help Implement Creative Payment Reform Solutions
As physicians consume 25% of every health care dollar (not to mention much of the 31% that hospitals consume) I think the "creative solution" is to make less (or better yet shift from costly specialists to less costly PCP's)
6) Talk About Reform With Patients
In one sense noble; in another sense creepy. It sounds kind of like "patients need to understand!". If patient have more "skin in the game" they will quickly understand the need for reform. What did LBJ say "When you have them by the b**ls, their hearts and minds will soon follow"
7) Minimize Conflicts of Interest
Again, people don't naturally reduce their income so this seems more like a system change than an individual physician change.
I do agree with your overall sentiment that physicians needed a better "system-sense" but that's a long term training issue
All good points. Thanks c3.
2 comments:
‘Even though the confidence intervals for overlapped, the authors based their conclusions on p values. This is outside the DMCB's statistical expertise, but suspects this has something to do with non-normal distributions involving odds ratios. That being said, the conclusion that the complication rates for more complex surgery are lower may be not quite correct.’
Confidence intervals and p-values are closely related in terms of construction. My suspicion, as a statistician, is that the authors used confidence intervals when looking at one-sample outcomes due to their increased interpretability (confidence intervals inform readers as to both estimated effect size and the precision of that estimate). The p-values were used when comparing differences between outcomes because the authors used two-sample statistical techniques, which offer more power than confidence intervals computed from one-sample techniques.
It is possible for two 95% confidence intervals computed from individual samples to overlap, while having sufficient evidence to conclude the population parameters differ at the .05 level of significance, because the techniques used in the two-sample case combine the information from both samples.
Even though the confidence intervals for overlapped, the authors based their conclusions on p values. This is outside the DMCB's statistical expertise, but suspects this has something to do with non-normal distributions involving odds ratios
I didn't enjoy my epi. stats course.
Post a Comment