Questioning Medicine

Be careful, most observational data have a large amount of confounders not accounted for and even when accounted for you can never account for all the confounders

https://www.ahajournals.org/doi/10.1161/CIR.0000000000001251. Diagnosis:To establish a PAD diagnosis, the resting ankle–brachial index (ABI) remains the initial test of choice in patients with suggestive history or exam findings. The ABI result should be reported as normal (1–1.4), borderline (0.91–0.99), abnormal (≤0.9) or noncompressible (>1.4). TREATMET Low-dose rivaroxaban 2.5mg BID, in addition to daily aspirin, is now recommended to decrease the risk for major adverse cardiovascular events (MACEs) and major adverse limb events in patients with symptomatic PAD who are not at increased bleeding risk. This is based on the COMPASS trial Cardiovascular Outcomes for People Using Anticoagulation Strategies - American College of Cardiology (acc.org) and as a reminder inclusion criteria was “Atherosclerosis in ≥2 vascular beds or two additional risk factors (current smoking, diabetes, renal insufficiency, heart failure, or nonlacunar ischemic stroke ≥1 month)” In pts with symptomatic PAD—single antiplatelet w

https://jamanetwork.com/journals/jama/article-abstract/2818624Conclusion--“Cardiovascular biomarkers were strongly associated with fatal and nonfatal cardiovascular events and mortality. The addition of biomarkers to established risk factors led to only a small improvement in risk prediction metrics for atherosclerotic cardiovascular disease, but was more favorable for heart failure and mortality.” For 10-year incident atherosclerotic cardiovascular disease in younger people (aged <65 years), the combination of high-sensitivity cardiac troponin I, N-terminal pro-B-type natriuretic peptide, and high-sensitivity C-reactive protein resulted in a C statistic improvement from 0.812 (95% CI, 0.8021-0.8208) to 0.8194 (95% CI, 0.8089-0.8277)  So this paper is saying look “Cardiovascular biomarkers were strongly associated with fatal and nonfatal cardiovascular events and mortality.” Not wrong these labs improve outcomes but when you look at the c stats we go from 0.81 to 0.82… remember as we talked about yesterday

What are c statisticsC-statistic gives the probability a randomly selected patient who experienced an event (e.g. a disease or condition) had a higher risk score than a patient who had not experienced the event.Obviously there are people with low scores that still have events and people with high scores that never have events but the goal is the decision score gives us an idea of who is most likely. ·       The idea or educated estimate can be turned into a C score-  and c scores are kind of like grades== a score of 1 is absolue perfect model it means the model perfectly predicts those group members who will experience a certain outcome and those who will not.·       But we know in medicine that isn’t possible  IF ·       A value of 0.5 means that the model is no better than predicting an outcome than random chance.·       Values over 0.7 indicate a good model.·       Values over 0.8 indicate a strong model.  

https://pubmed.ncbi.nlm.nih.gov/38598573/In fully vaccinated adults with a risk factor or unvaccinated patients without a risk factor who have symptomatic COVID-19, does paxlovid--nirmatrelvir-ritonavir reduce the duration of symptoms or the likelihood of hospitalization?   Nirmatrelvir-ritonavir (Paxlovid) was shown in its initial randomized trial to reduce hospitalization and death in unvaccinated adults with at least one risk factor for severe disease when the ancestral variant of SARS-CoV-2 was predominant.But it is important that drugs be evaluated in the correct target population patients who have been vaccinated or have the Omicron variant.This industry-sponsored study enrolled 2 groups of patients: (1) fully vaccinated adults with symptomatic, confirmed infection with SARS-CoV-2 and at least one risk factor for severe disease,(2) unvaccinated adults with a symptomatic infection but no risk factorsThe onset of symptoms was within the past 5 days. Patients (N = 1296) were randomized to receive the stand

https://www.nejm.org/doi/10.1056/NEJMoa2314063Randomized trials have shown the benefit of endovascular thrombectomy in patients with acute stroke due to large-artery occlusion in the anterior circulation and a large baseline infarct (core) In these trials,1-5 a large core was defined by an ASPECTS value of 5 or less, but because of concerns about the deleterious effects associated with the reperfusion of large infarcts,8 patients with the largest infarcts (ASPECTS value, 0 or 1) were excluded from enrollment Now, researchers have compared EVT plus medical therapy to medical therapy alone in patients who could be treated within 6.5 hours of stroke onset and had a large amount of ischemic tissue The primary outcome was the modified Rankin scale (mRS) score and the major safety outcome was all-cause mortality, both at 90 days.  a 3-year period, 333 patients  cerebral vessel occlusion in the anterior circulation The median NIH Stroke Scale score was 21, and the median baseline infarct volume was 135 mL;  Median t

https://www.ahajournals.org/doi/full/10.1161/STROKEAHA.124.047438We studied the risk of IS and ICH in women using LG-IUDs—aka mirana-- compared to women not using hormonal contraceptives.  The commonly used combined hormonal contraceptives with progestins and ethinylestradiol are associated with an increased risk of ischemic stroke (IS).   In this Danish historical cohort study (2004-2021) we followed non-pregnant women (18-49 years) registering incident IS and ICH in relation to use of LG-IUD/non-use of hormonal contraceptives utilizing Danish high-quality registries with nation-wide coverage. A total of 1,681,611 non-pregnant women A total of 1,681,611 non-pregnant women The numbers get huge here however-- After adjustment incidence rate ratio for IS was 0.78 (CI: 0.70; 0.88), and for ICH it was 0.94 (CI: 0.69; 1.28)   Use of LG-IUD was associated with a 22% lower incidence rate of IS without raising incidence rate of ICH. I think IUD are unutilized and now we have data that says they have lowers rates of s

 A noninferiority trial is a type of randomized trial that aims to establish if treatmenet X effectiveness is not substantially less than the existing standard. Unlike superiority trials that are designed to show that one treatment is better than another, a non-inferiority trial is designed to show that a new treatment is ‘not unacceptably worse’ than the current standard therapy. You want to see if the drug you are testing is good enough compared to the standard. Say you wanted tacos from your favorite place across town cause they have the best tacos but there is another taco place that is across the street the tacos are not as good but they are pretty close and it is a 20 second walk not a 20 minute drive. You might quickly in your mind calculate the trade off and say well the place across the street is noninferior.

https://shmpublications.onlinelibrary.wiley.com/doi/abs/10.1002/jhm.13350Xu J et al. Hospital-associated venous thromboembolism prophylaxis use by risk assessment at a large integrated health care network in Northern California. J Hosp Med 2024 Jun; 19:449. Authors took 850,000 adult nonsurgical, non–intensive care unit (ICU) hospitalizations at 21 Kaiser Permanente hospitals in northern California, and did a retrospective study of inpatient pharmacologic VTE prophylaxis, investigators compared risk assessment by admitting physicians with risk assessment according to electronic health record (EHR)-The EHR used the Padua prediction score which basically ask yes or no questions like does the pt have active cancer, previous vte, reduce mobility, elderly age, heart or resp failure. All questions that could need a human to fill out but also with could AI or HER should be answered without humans doing anything. In 82% of 850,000 adult nonsurgical, non–intensive care unit (ICU) hospitalizations, the EHR categorized

https://www.bmj.com/content/385/bmj-2024-079329Overprescription of antibiotics by primary care clinicians is a major modifiable driver of antibiotic resistance. Evidence suggests that peer-comparison feedback can reduce antibiotic overprescription, but the optimal content and delivery of such feedback is unclear   Researchers randomized 5000 family physicians in Ontario, Canada, to receive either mailed feedback (with data on individual prescribing rates compared with peers' prescribing rates, plus educational information on optimal prescribing) or no mailed feedback (control group).Clinicians in the intervention group were randomized further to (a) receiving personalized prescribing data that were adjusted, versus not adjusted, for case mix, and (b) receiving information on potential harms of antibiotics, versus not receiving such information.Compared with controls, the intervention group had significantly lower mean rates of overall antibiotic prescribing (59 vs. 56 prescriptions per 1000 patient visits; re

Kityo C et al. Switch to long-acting cabotegravir and rilpivirine in virologically suppressed adults with HIV in Africa (CARES): Week 48 results from a randomised, multicentre, open-label, non-inferiority trial. Lancet Infect Dis 2024 May 28; [e-pub]. (https://doi.org/10.1016/S1473-3099(24)00289-5)307

95% CI does NOT mean there is a 95% chance the true value is in the interval. It means that if you continually resample, 95% of your confidence intervals (which will vary every time you resample), will contain the true or real value of the intervention. 

https://academic.oup.com/jcem/advance-article/doi/10.1210/clinem/dgae290/7685305?login=trueChildren aged 1-18 years to prevent rickets and to potentially lower the risk for respiratory tract infectionsPregnant people to lower the risk for maternal and fetal or neonatal complicatioAdults older than 75 years to lower the risk for mortalityAdults with prediabetes to lower the risk for type 2 diabetes

 What is the power of a study—what is the power calculation? Often say it is the number of people in a study. Power is the probably to correctly reject the null hypothesis. Said differently power is the probability we will correctly get a small pvalue But a power calculation is slightly more than that it is the number of people required to adequetyly statistically calculate the number of people that would be necessary to reject your null hypothesis of no difference when there actually is a difference. Lets say we have a blood pressure drug and we want to see if it works Well if you only enroll a small number of people you might not be able to tell a difference between those in the active and those in the control arm Remember depending on baseline bp you might only see a 4-5mmg hg improvement in your blood pressure and in almost every trial I have ever seen even those individuals that get the placebo have a 1-2mm hg improvement in their bp so with large confidence intervals that are likely to overlap with just

Question  What is the strength of association between surrogate markers used as primary end points in clinical trials to support Food and Drug Administration (FDA) approval of drugs treating nononcologic chronic diseases and clinical outcomes?   often surrogate markers are used as primary end points in clinical trials to support FDA approval of drugs I get it  Surrogate markers offer the advantage of reducing the duration, size, and total cost of trials n 2018, the Food and Drug Administration (FDA) publicly released an Adult Surrogate Endpoint Table of more than 100 surrogate markers that may be used as primary end points in clinical trials that form the basis of traditional or accelerated approval of new drugs or biologics.   The authors evaluated Thirty-seven surrogate markers listed in FDA’s table of markers that can be used as primary end points in clinical trials across 32 unique nononcologic chronic diseases. Most surrogate markers used as primary end points in clinical trials to support FDA approval o

The incidence of new-onset diabetes was basically the same but statistically significantly higher for those individuals on low-to-moderate–intensity statins compared with placebo 1.2 vs 1.3% annually which is a very small difference. But with  high-intensity statins compared with placebo (4.8% vs. 3.5% annually) Among patients with known diabetes at baseline, glycemia worsened slightly with statin therapy compared with placebo Here is the problem- diabetes is a number—a surrogate if you will. Statins fix a surrogate but have been proven to improve patient orientated outcomeshttps://www.clinicalkey.com/#!/content/playContent/1-s2.0-S2213858724000408?returnurl=https:%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS2213858724000408%3Fshowall%3Dtrue&referrer=https:%2F%2Fwww.jwatch.org%2F

At change in c stats of 0.007 or 0.0009 is not a meaningful change so I cant say we should use this over the PCE—yes this new calculator has the benefit of removal of race, and the use race-based algorithms. We don’t know that this leads to better outcomes—is the the race algorithms that lead to worse outcomes or was it access to care or is it some other factor we don’t know yet. I think this is worth nothing and if you want to switch you certainly can but if your goal is a calculator to be used to detect primary CAD or to use in your primary CAD population EITHER seems to be just fine at this time.https://pubmed.ncbi.nlm.nih.gov/37947085/

The problems with observation data is real—  randomized trial, U.K. researchers identified 519 patients (mean age, 68) with mostly moderate COPD (mean forced expiratory volume in 1 second [FEV1], 50%), ≥2 exacerbations during the previous year, and no cardiovascular (CV) indications for β-blockers.  Patients were randomized to receive the cardioselective β-blocker bisoprolol (initially 1.25 mg daily, titrated to 5 mg if tolerated) or placebo. At 1 year, no significant differences were noted between groups in incidence of COPD exacerbations or in other important benefits or harms.Cardioselective β-blockers remain appropriate for COPD patients who have valid cardiovascular indications for their use, but taken these two studies together suggests that COPD patients without such indications should avoid bblockers—even cardio selective beta blockershttps://jamanetwork.com/journals/jama/article-abstract/2819083

When you want to do an RCT but you realize there might be some cross contamination so instead of randomizing individuals you randomize cluters or groups Lets say you want to do fluids--- what is better LR or NS instead of doing individuals and room you just put whole hospitals in either LR or NS Or you want to test if mask work for some virus—you would say this city gets all the mask in the world and this city gets zero mask…you know that If you just gave some individuals in the city a mask then maybe someone who give their extra mask to their friends and you would have cross contamination meaning that that individuals in the group end up getting the intervention you are trying to test. Cluster RCT are GREAT for logistics and trying to figure out logistically can something or does something actually work However cluster RCT can be harder to analyze because the clusters may not always have the same exposures or the confounders might not be equal between clusters In our mask example that we just talked about le

1° outcome: hemostatic efficacy = which was defined as;(A) Hematoma expansion ≤ 35% at 12h(B) increase in NIHSS ≤ 7 at 12h(C) No rescue therapy 3-12h    Results hemostatic efficacy: 76.7% andexanet vs 64.6% usual care    30 day mortality: no difference30 day Modified Rankin Score ≤3: no differenceThrombotic events: almost x2 with andexanet 10.3% vs 5.6% - statistically significant. Most were strokes and MI - not trivial. The question you should ask: How does effective is hemostatic efficacy as a marker for patient outcomes?? Yes, we don’t want the brain bleed to get bigger but patients don’t care if the bleed gets bigger if they still die or if they still are in a coma for the rest of their life. We know from warfarin that increase in hematoma expansion leads to worse outcomes (https://pubmed.ncbi.nlm.nih.gov/21346218/). We don’t have clear data on this for the DOACs. HOWEVER, just because an expanding hematoma leads to a bad outcome that does not mean that giving a medication to decrease hematoma expansion l