- Joined
- Dec 2, 2010
- Messages
- 154
- Reaction score
- 90
The below is a claim and then a response by a rather intelligent physician, who also formerly worked in the health insurance industry named "vagusnight". All credit to him/her.
--------------------------------------------------------------------------------------------------------------------------
Claim: Patients do receive the same level of care. Physicians and NPs have equivalent outcomes among their primary care patients: http://jama.jamanetwork.com/article.aspx?articleid=192259 http://www.ncbi.nlm.nih.gov/pubmed/24586577 http://www.ncbi.nlm.nih.gov/pubmed/24194798 etc...
--------------------------------------------------------------------------------------------------------------------------
Response (vagusnight): These studies almost universally have the same underlying theme: patients are pre-screened for conditions that absolutely require medical intervention, and the remaining randomization group tends to be things that are benign, self-limiting, or lifestyle issues that rarely improve because patients don't act on it regardless. Let's see if these 3 are an exception:
Mundinger (2000): Doesn't provide explicit details of what "screening criteria" were (earlier methods statement says they are invited to apply if they don't currently have a PCP and plan to be in the area for 6 mos., but doesn't explicitly restrict it to that. Study loses a whopping 40% of patients at the recruitment stage - this doesn't openly bias in one way or another, but it fundamentally puts the study validity into question w/o explicit inclusion criteria. Tracks primary care for a whopping total of 6 months post-ED and UC visit, a very short period of time for chronic conditions - especially post-ED visit (their recruitment center), which would imply a strong regression to the mean for all involved parties that could easily mask inter-group differences, biasing us towards the null. Doesn't distinguish between patients based on presenting condition (saying that patients with asthma look the same in 6 months is different depending on whether they presented to the ED for uncontrolled asthma, or presented to the ED for a toe fx.) Notably, and not surprisingly coming from a bunch of DrPH authors, they focus on diseases where dx is a cinch and treatment is absurdly standardized - in other words, with a study biased heavily towards the null.
Gonzalez (2014): Meta-analysis of small, biased studies from other countries with different training regimes and different distributions of clinical authority. Fundamentally lacking in external validity; I don't need to dig into them to look for internal validity.
Health Quality Ontario (2013): "Due to clinical heterogeneity in the study populations evaluated, and differences in provider roles and characteristics, the pooling of outcomes was thought to be inappropriate and a meta-analysis was not conducted. Outcomes were summarized descriptively, with significance accepted at P < 0.05." Oh, good, an editorial review masquerading as a meta-analysis. Glad we only had to get half-way through the methods section before that became clear. Well, let's see how many papers they rev- oh, 6. Well, I'm sure they at least have a few new papers about how well nurses substitute for doc - oh, only two. And one of them is the same Mundinger paper I tore into above. So they actually only have one additional paper to add to this discussion, Lenz 2002. Another reasonably stable chronic disease study looking at short time periods (6 mos again), though this time looking at only about a hundred and twen-
Oh. It's actually a post-hoc sub-group analysis of the Mundinger data. It's not actually new data at all. I guess I don't care about Lenz 2002, either.
I know people seem brusque and over-generalizing when they say things like "all those studies saying mid-levels are comparable to docs are just garbage," which is why I put in the time to dig into them one by one. None the less, those people are right: all those studies saying MLPs are comparable to MDs are just garbage.
--------------------------------------------------------------------------------------------------------------------------
Claim: Patients do receive the same level of care. Physicians and NPs have equivalent outcomes among their primary care patients: http://jama.jamanetwork.com/article.aspx?articleid=192259 http://www.ncbi.nlm.nih.gov/pubmed/24586577 http://www.ncbi.nlm.nih.gov/pubmed/24194798 etc...
--------------------------------------------------------------------------------------------------------------------------
Response (vagusnight): These studies almost universally have the same underlying theme: patients are pre-screened for conditions that absolutely require medical intervention, and the remaining randomization group tends to be things that are benign, self-limiting, or lifestyle issues that rarely improve because patients don't act on it regardless. Let's see if these 3 are an exception:
Mundinger (2000): Doesn't provide explicit details of what "screening criteria" were (earlier methods statement says they are invited to apply if they don't currently have a PCP and plan to be in the area for 6 mos., but doesn't explicitly restrict it to that. Study loses a whopping 40% of patients at the recruitment stage - this doesn't openly bias in one way or another, but it fundamentally puts the study validity into question w/o explicit inclusion criteria. Tracks primary care for a whopping total of 6 months post-ED and UC visit, a very short period of time for chronic conditions - especially post-ED visit (their recruitment center), which would imply a strong regression to the mean for all involved parties that could easily mask inter-group differences, biasing us towards the null. Doesn't distinguish between patients based on presenting condition (saying that patients with asthma look the same in 6 months is different depending on whether they presented to the ED for uncontrolled asthma, or presented to the ED for a toe fx.) Notably, and not surprisingly coming from a bunch of DrPH authors, they focus on diseases where dx is a cinch and treatment is absurdly standardized - in other words, with a study biased heavily towards the null.
Gonzalez (2014): Meta-analysis of small, biased studies from other countries with different training regimes and different distributions of clinical authority. Fundamentally lacking in external validity; I don't need to dig into them to look for internal validity.
Health Quality Ontario (2013): "Due to clinical heterogeneity in the study populations evaluated, and differences in provider roles and characteristics, the pooling of outcomes was thought to be inappropriate and a meta-analysis was not conducted. Outcomes were summarized descriptively, with significance accepted at P < 0.05." Oh, good, an editorial review masquerading as a meta-analysis. Glad we only had to get half-way through the methods section before that became clear. Well, let's see how many papers they rev- oh, 6. Well, I'm sure they at least have a few new papers about how well nurses substitute for doc - oh, only two. And one of them is the same Mundinger paper I tore into above. So they actually only have one additional paper to add to this discussion, Lenz 2002. Another reasonably stable chronic disease study looking at short time periods (6 mos again), though this time looking at only about a hundred and twen-
Oh. It's actually a post-hoc sub-group analysis of the Mundinger data. It's not actually new data at all. I guess I don't care about Lenz 2002, either.
I know people seem brusque and over-generalizing when they say things like "all those studies saying mid-levels are comparable to docs are just garbage," which is why I put in the time to dig into them one by one. None the less, those people are right: all those studies saying MLPs are comparable to MDs are just garbage.
