Plasma Volume Estimates and Mortality

Measuring plasma volume is quite difficult and is an involved process. Therefore, equations have been developed that estimated plasma volume by some easily measured variables. One such equation is Straus formula. Using this formula, Marawan et al found that estimated plasma volume is not only related all-cause mortality but also to cardiovascular, cancer, and other-cause mortality.

These findings beg the question about the underlying mechanism of this association. It is difficult to say how this works except that perhaps one of the common final pathway of most diseases is to increase plasma volume.

Socioeconomic Status and Readmission Penalties

CMS will be including socioeconomic status when penalizing hospitals for excess readmissions within 30-days of discharge from hospital. the socioeconomic status will be determined by calculating the percentage of population seen at a hospital that are dual eligible (both Medicaid and Medicare).  See here for the details.

There is hope that this adjustment will allow hospitals that take care of very sick patients but also socioeconomically disadvantage populations will find a more even playing field when competing with hospitals that serve relatively wealthier populations.

It is a no brainer that patients with strong support systems, easy and quick access to their primary care provider, and high health literacy are less likely to be readmitted as compared to patients who lack these factors. Often, poor support system, poverty, lack of primary care access, and limited health literacy go hand in hand. These disadvantaged patients are most difficult to treat and are higher risk of disease-related and treatment-related complications.

Now Maddox et al has calculated which hospitals are likely to benefit and how large the benefit is going to be. They found that hospitals in the lowest quintile of dual enrollment (that is hospitals in the relatively wealthier neighborhoods) saw an increase of $12.3 million in penalties, while those in the highest quintile of dual enrollment (serving disadvantaged populations) saw a decrease of $22.4 million. Large hospitals, teaching hospitals, hospitals in the most disadvantaged neighborhoods, and those with the highest proportion of beneficiaries with disabilities were markedly more likely to see a reduction in penalties, as were hospitals in states with higher Medicaid eligibility cutoffs.

Shared Decision Making in Medicine

I was reading this ‘viewpoint’ in JAMA; an interesting read. I try to practice shared decision making as much as possible, perhaps more than what my patients would like.

However, I do feel that there is lack of data supporting shared decision making or ‘informed decision making’. I would perhaps define ‘informed decision making’ as decision making process in which patients are informed of their options and then given a recommendation by their treating physicians. Of course, patients have a right to say no to the recommended option, give reasons for refusing the recommended option, and then give their preferred option to the physicians.

Having said that, I am unaware of good data supporting any decision making strategy. There are, of course, tons of arguments supporting why shared decision making is good. But when, in medicine, we started depending on arguments without trying to develop evidence?

To develop evidence, first we have to define what is it that we want to achieve (an outcome). Some may argue that ‘shared decision making’ is in itself an outcome. I am not sure how this can be an outcome of choice. Perhaps a better outcome could be one that is patient-focused, such as mortality, morbidity, readmissions, quality of life etc. Things that matter are more meaningful. Or outcome that is physician-focused: such as number of law-suits by patients (or on behalf of patients). But there need to be reasonable outcome(s).

Then, we need to examine how shared-decision making affects those outcomes. My worry is that we are adding ‘must-do’ things in our workflow without determining their efficacy.

One way I look at the seriousness, or lack of it, by the proponents of a given ‘thought’ or ‘idea’ is their lack of desire to investigate their proposed remedy using scientific methods. We don’t accept a medicine, procedure, or treatment without strong evidence. Why we should accept other things using a lesser proof of evidence (unless we think it is not as important)?

One area where we find such a lack of seriousness in developing evidence is medical education itself. Even when we occasionally study a medical education intervention correctly, we find that our thinking was flawed. While we may claim that we understand the complexity of human behavior, we are often wrong. I doubt that we are right when it comes to shared decision making. When properly studied, I doubt we will find any effect on any important outcome.

Physician Burnout is Rising

Physicians have higher levels of burnout than the general US population. There are several proposed factors for this higher prevalence although none have been proven beyond doubt.

In addition to higher prevalence, the prevalence is increasing raising concerns for the health of some of the highly educated and skilled professionals in our society. A recent study found that the prevalence of burnout in a large academic medical center faculty increased from 40.6% to 45.6%. Both domains of burnout increased; exhaustion increased from 52.9% to 57.7% while depersonalization increased from 44.8% to 51.1%.

Other things that changed were that fewer physicians in 2017 felt that they had control over their schedules than in 2014 (71.6% vs. 64.3%), felt that they had could impact decision making that affects day to day practice (58.7% vs. 55.4%)  and were satisfied with their workload (55% vs. 50.5%).

Authors also validated a finding from prior studies that early-career physicians are more susceptible to burnout than mid-career physicians. However, they did not find a difference in burnout by gender or by race.

Readmission Rate and Insurance Type of the Patient

One of the programs instituted through the Affordable Care Act (ACA), also known as Obamacare, is the Hospital Readmissions Reduction Program (HRRP). This program focuses on the readmission of the patients within 30-days of discharge. Hospitals that perform worse than the national average are penalized up to 3% of their Medicare payments. While the program targets Medicare beneficiaries, it is likely that changes hospitals make to reduce readmissions will benefit all patients.

An article published in Health Affairs examines this question. Authors find that the readmission rates declined after the HRRP went into effect (not surprising as it has been shown by us and many others). In addition, they also report that readmissions declined for both Medicare and Medicaid patients. However, readmission rate of Medicaid patients remained higher than Medicare patients. Readmission rate was lowest for patients who had insurance throughout the study period and declined at a similar rate compared to others.

This study highlights an important but often neglected area when examining readmissions. Readmissions depend on three main factors:

1) Hospital Discharge Practices: These are factors that are under the control of the hospital (physicians and hospital employees). These include adequate discharge instructions, adequate hand-offs to the outpatient provider, adequate education about patient illness during hospitalization, making followup appointments, etc.

2) Patient factors: These includes the severity of the underlying illness of the patient. Some patients are far sicker and even when they are discharged from hospitals, they are just beyond the edge of the need for hospitalization. A small change in patient’s condition leads to a visit (and admission) back to the hospital.

3) Social support factors: These include support from family, friends, community, and other resources.

Of these three, very little attention has been paid to the third main factor, social support. Patients with private insurance are likely to have more resources and stronger support. On the other hand, patients on Medicaid are likely to be poor with an inadequate social support system. I believe that we will start hitting the wall with regard to the readmission rate reduction and the rate of decline in readmissions will start slowing down (if it is already not slowing). Ultimately, as a society, we will have to focus on the social support aspect of readmissions to keep patients healthy and out of the hospitals.

Abstain from Alcohol to Decrease Risk of Atrial Fibrillation

Here is a press release about a study that was presented by has not yet been published. Perhaps the best study design for evaluating the benefits/harms of alcohol.

Lead Exposure in Childhood can lead to Personality Changes

Here is the link to a simplified version

Readmissions and Mortality–Any Difference by Race?

Since the introduction of Hospital Readmission Reduction Program by the CMS, there has been a significant decrease in readmissions for all targetted diseases. However, studies have raised concern that although readmissions are decreasing, there may be an increase in mortality.

A recent study showed that there is no difference in mortality between white and black patients. Authors used interrupted time-series analysis. I enjoyed statistical modelling but wonder if the within-hospital and between hosptial effects were properly accounted for. In other words, the assumption in their modeling is that both effects are the same (unlikely to be true). It is possible that the within and between effects are different and may shed a better light on what is happening at individual hospital level and what is happening across hospitals. Having said that, it is an interesting study with important implications.

Are Machine Learning Tools Better than Standard Tools in Predicting Readmissions

Just saw this study where authors compared standard readmission tools (HOSPITAL score, modified LACE score, and Maxim/RightCare score) with a model developed using machine learning. Authors found that machine learning score (they called it Baltimore Score or B-score) performed much better than standard tools.

While I agree that machine learning tools will likely outperform standard methods. Standard methods are quite a bit of oversimplification of the real life, machine learning tools less so. However, I doubt that authors have got their model right. Two reasons: One, their sample size is relatively small. Two, they have not validated their tool in a new dataset.

Social risk factors adjustment for readmission penalties

I have shown, in my previous work, that social risk factors affect hospital performance. CMS, at th eurging of Congress, will be including including social risk factors in their patient risk models. Here is an interesting paper that examined retrospectively the effect of adjustment for social risk factors on readmission penalties and found that the penalties will drop by almost half for safety-net hospitals. I am certain that these adjustment will bring more fairness in hospital comparisons and will decrease the amount of penalty these safety-net hospitals have to face. 

Obesity and Hospital Readmissions

The link between obesity and readmissions is not as clear cut, despite the fact that almost 40% of older adults in the US are obese. In this relatively large study examining the readmission rates of obese nursinghome patients, authors found that the relationship is perhaps complex and islikely affected by age and co-morbid conditions.

Rebate on Prescription Drugs

Illuminating article; must read. Here are two sentences that capture the essence of this issue: “In 2016, leading products in these classes had total pre-rebate spending in Part D of $2.5 billion (Lantus Solostar) and $4.4 billion (Harvoni). For an individual patient, this would reduce list prices on which their deductibles and co-insurance are calculated from $417 to $142 for patients filling a 1-month supply for Lantus Solostar and from $92 648 to $35 206 for patients filling an 84-day course of treatment with Harvoni, assuming class-level rebates applied to these products.”

Failure of Patient-centerd Transitions of Care Program to Improve Outcomes in Heart Failure Patients

Improving patient care through focus on transitions of care is hard! There are very few studies that have evaluated strategies that work in a randomized controlled trial (RCT) fashion and none has been able to show any benefit. Here is a large RCT that failed to show a benefit.

What is that works? And works consistently?

Number Needed to Treat

Number needed to treat is an important concept as it simplifies the communication of the effect size from clinical studies. It tells us the number of patients needed to be treated to get the desired outcomes (or an undesireable adverse event). Here is a much longer description from JAMA.

Physician Organizations’ contribution to political Candidates and NRA ratings

Must read article in JAMA. Here is a graph that says is all; full article is here.

Quick Overview of Mediation Analysis

JAMA recently published this overview of Mediation Analysis; often a misunderstood statistical method (and some may argue a method for causal modeling), mediation analysis is a powerful technique that can help to delineate biological or psychological mechanisms.

Physicians Support Tighter FDA Control on New Drug Approval

An interesting research letter published in JAMA Internal Medicine reports that 80% of physicians agree with the strict FDA approval process to protect public from ineffective or dangerous drugs. Further, 60% thought that FDA should not allow off-label promotion of drugs to physicians.

Genistein, a Phytoestrogen, and Mortality

Several studies in medical literature and countless lay-media articles promote the benefits of phytoestrogens. In this large study, using data from NHANES, Marcelo and colleagues showed that that high levels of genistein in urine are associated with lower mortality. In particular, it decreases mortality from all causes.

Apixaban in ESRD and Atrial Fibrillation

Patients with ESRD (end-stage renal disease) are usually prescribed coumadin when they need anticoagulation; the reason being that we know very little (or not at all) about the effect of new oral anticoagulants in patients with ESRD.

In this study, Siontis et al performed a retrospective cohort study of Medicare beneficiaries with ESRD and atrial fibrillation, the later an indication for anticoagulation. Outcomes were compared between patients who were treated with apixaban versus patients who were treated with warfarin. Authors examined the survival difference between the two groups for stroke or systemic embolism, major bleeding, gastrointestinal bleeding, intracranial bleeding, and death using Kaplan–Meier analyses and hazard ratios (HRs) and 95% CIs were obtained using Cox regression analyses.

Authors found that there was no difference in stroke and systemic embolism between the two groups. However, the group taking apixaban had 28% lower risk of major bleed.

While this is retrospective data, and clinical trial data is lacking, this study will support the use of new oral anticoagulants in ESRD patients.

Is Science Getting Less Bang for its Buck?

Here is an interesting article published in The Atlantic which claims that the productivity in science is declining. The investment in science is increasing but the pace of novel or truly important discoveries is decreasing. They give various reasons; I have few additional thoughts that I assume authors thought of but were not included in this piece.

1. The age of scientist at discovery is increasing due to the large volume of knowledge we ‘force’ young scientist to learn. We have formalized ‘learning’ to an extent that we feel it is obligatory for anyone to ‘know’ it all before moving forward. Just as an example, the ‘required’ courses to take are often not directly relevant to the learner’s interest or the to the topic. WE have added those without any scientific evidence that such additional courses help with discovery. We preach evidence, we prefer not to teach by science. And when we try to accumulate evidence, we do it using irrelevant outcomes. This is the problem with education in general and not with science education.

2. Truly novel discoveries happen in spurts. Discovery of zero was truly important, one may argue more important than general relativity. Perhaps discovery of algebra and calculus were also more important, and fundamental, than general relativity. But these discoveries are not ‘novel’ anymore for us. I believe new discoveries happen after we have utilized the older discoveries to close to the fullest extent. Imagine a new discovery as a new house; once we build a new house, we start filling it, decorating it, accumulating useful (and some junk) ‘stuff’. This is perhaps what happens with truly important discoveries. A truly important discovery is followed by many small discoveries over next many years until the potential of new discovery has been fully realized and then another new discovery happens.