Zackary Berger’s journal reviews—28 August 2018

This week we introduce the first of our three new columnists who will be writing our weekly research reviews

Zackary Berger is an associate professor at Johns Hopkins School of Medicine in the division of general internal medicine, and core faculty in the Johns Hopkins Berman Institute of Bioethics. His research focuses on shared decision making, patient-centered care, and health justice. He is also a practising physician at a free health clinic for undocumented, predominantly Spanish-speaking, immigrants.


Alcohol use and global burden

Two different purposes underlie this study: to quantify the health effect of alcohol use; and to determine whether in fact there is some small amount of alcohol use which is beneficial.

The authors amassed a large number of data sources and conducted 21 meta-analyses to estimate the influence of alcohol on health outcomes. A limitation of this approach is the issue of self-reported alcohol consumption; further, accruing a large number of studies which are potentially biased does not remove the bias in these individual studies. That alcohol confers an important burden of disease is indisputable, and the associations between alcohol and these individual outcomes is an important product of this work—whether causality can be inferred is for epidemiologists, physicians, and patients to discuss.

Whether “the level of consumption that minimizes health loss is zero,” depends (as has already been pointed out by many perceptive readers) on how to read figure 5 in the paper, which shows that at one standard drink per day, the relative risk for all attributable causes was 1. Whether emphasizing zero alcohol exposure as a public health priority above, say, treatment of those with alcohol use disorders remains a difficult question. Whether a drink a day leads to a relative risk of 1, or very near 1, might lack relevance compared to higher-dose exposures.

Epinephrine in out-of-hospital cardiac arrest

Researchers in the UK undertook an RCT of epinephrine in cardiac arrest, conducted in five NHS ambulance services. Does epinephrine improve survival at 30 days over cardiopulmonary resuscitation and defibrillation alone, when the former are unsuccessful? The short answer is yes, with an absolute difference in 30-day survival of 0.8%—that is, a number needed to treat of 125, compared to an NNT of 15 for CPR performed by a bystander and 5 for early defibrillation. A logistical quirk of this study is that family members or patients were asked for consent after the emergency had passed.

The longer answer, however, is that survivors in the epinephrine group had twice the rate of severe neurologic impairment. As the authors point out in their discussion, every intervention is associated with a balance of risks and benefits. Whether the benefit of increased chance of survival is worth the risk of neurologic impairment (through what mechanism, we can only speculate—perhaps involving the adrenergic receptors and impaired cerebellar blood flow), is something that can be answered only by further studies providing risk stratification. In the moment of cardiac arrest, given these uncertainties, such a decision seems a Sisyphean task for families, and perhaps the system should give a nudge towards a justified default.

Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes

Can the excess risk of heart and vascular disease in patients with type 2 diabetes be eliminated through control of risk factors? While multiple previous randomized controlled trials have considered the effect of single risk factor modification on cardiovascular mortality and morbidity, the research team from Sweden used a retrospective cohort design to assess whether control of multiple risk factors could eliminate the excess risk. The risk factors are the familiar usual suspects: hemoglobin A1C, low-density lipoprotein, smoking, and blood pressure, together with albuminuria. The excess risk of death, myocardial infarction, stroke, or hospitalization for heart failure decreased stepwise for each risk factor within the target range. The hazard ratio for all-cause death among those with diabetes with the five risk factors in the target range was 1.06 (95% CI, 1.00 to 1.12). Smoking was the strongest predictor of death among those with type 2 diabetes, and hemoglobin A1C was the strongest predictor of stroke and myocardial infarction.

Association between traumatic brain injury and risk of suicide

Like other Scandinavian nations, Denmark registers every individual with a unique identifier, which makes studies like this one possible, or at least more feasible. The researchers assessed the association between “medical contacts” (inpatient, outpatient, or emergency visits) for traumatic brain injury (TBI) as recorded in the national patient register, and suicide as recorded in the Danish cause of death register. Previous studies have examined this association, but few at this scale or with the information necessary to control for a range of variables including other types of fractures besides cranial fracture, psychiatric illness, epilepsy, and a range of demographic and socioeconomic variables. (I did not see substance use disorder as a named covariate, though this might have been included under psychiatric diagnoses.)

Unsurprisingly, TBI was associated with an approximately 2-fold higher risk of suicide. Other findings in this study to do with severity of TBI and higher suicide rate among those experiencing TBI in young adulthood buttress a dose-response relationship. However, the association between suicide and TBI is nothing as simple as direct causality and involves psychiatric illness predisposing to TBI, pre-existing systemic brain disorders, and numerous ways in which psychiatric illness can affect access to medical care. The association laid out here is a potential foundation for more detailed work mapping the highways and byways of these factors.

The safety of discharging select patients directly home from ICU

In a retrospective population-based cohort study in nine hospitals in Alberta, Canada, Stelfox and colleagues assessed whether discharging patients from a medical ICU directly home, as opposed to transfer to a hospital ward, was associated with readmission within 30 days (the primary outcome), emergency room visit within 30 days, or death within 1 year.

Patients discharged directly home were younger, more likely to have been admitted with overdose, substance withdrawal, seizures, or metabolic coma, to have a lower severity of acute illness on admission to the ICU, and have been mechanically ventilated for less than 48 hours. There were no significant differences in the outcomes assessed. This study is essentially proof of a concept which has already, it seems, been put the test over a number of years (at least in Canada). Are there other diagnoses for which this strategy might work, and which subgroups of patients might it be suited for? Are there other outcomes—ICU-related psychosis or psychological trauma, for example, or physical mobility—which should be examined in future studies? Are these results generalizable to other locales and patient populations? All of these questions are worth examining.

Shared decision making for lung cancer screening

When the US Center for Medicare and Medicaid Services decided in 2015 to cover lung cancer screening via yearly low-dose CT scan, it mandated that shared decision making or informed decision making be part of screening discussions. However, this mandate came without either funding or teeth. Thus I can’t say it is surprising that Brenner and colleagues study found little evidence of shared decision making in primary care or pulmonology visits with patients who might be eligible for LCS. This was a qualitative study of 14 transcripts from private practice outpatient encounters. In fact, going by the validated OPTION instrument, which the authors used to assess the presence of shared decision making behaviours, the physicians’ performance was dismal, with an average score on the OPTION12 of 8 out of 100 (a systematic review found the mean among practising physicians to be 23). None mentioned different options for screening lung cancer (e.g., CT vs watchful waiting or no screening).

Why is this? One possibility is that US physicians aren’t very good at SDM no matter what decision is being discussed, and they need training to become expert. This seems most likely. In fact, lung cancer screening (possible small benefits, likely harms ranging from small to large, on the whole a preference-sensitive decision) is like many population-health interventions which physicians feel pressured to recommend. Everything (still!) seems to tend against SDM, the EHR and billing included. Incentives need to be pitched in the opposite direction.

HIV viral suppression trends among HIV-infected patients in the US

Since 1996, antiretroviral therapy has increased survival for people living with HIV (PLWH), but that discovery came too late for thousands, and those with HIV were ignored by governments and many practitioners. This cohort study sets out to estimate how frequently HIV is suppressed among PLWH in 8 well-resourced HIV clinics in the US—representing a patient population of approximately 32,000 out of 1.2 million PLWH in the US—what factors are associated with suppression, and who ART has yet to reach.

Among this group, viral suppression increased to 86% in 2015 from 32% in 1997. Being older and using an integrase-based regimen were associated with a lower odds of a detectable viral load. Black race was associated with a higher odds of detectable viral load—that is, lesser odds of suppression. The importance of this finding is not adequately conveyed by the term “health disparities,” which tends to treat race as a covariate rather than a reflection of socioeconomic structures. In their discussion, the authors suggest that decreased HIV suppression among black patients might be due to “substance use, adherence, or other factors, such as medical mistrust” (note that substance use and adherence were not associated with suppression in the total cohort). It is notable that systematic racism is still infrequently mentioned as such in the biomedical literature, but racial disparities are mentioned quite frequently.

Zackary Berger, is an associate professor at Johns Hopkins School of Medicine in the division of general internal medicine, and core faculty in the Johns Hopkins Berman Institute of Bioethics, both in Baltimore, Maryland. Zackary’s research focuses on shared decision making, patient-centered care, and health justice. He is also a practising physician at a free health clinic for undocumented, predominantly Spanish-speaking, immigrants.

Competing interests: None declared