Zackary Berger reviews the latest research from the top medical journals

Developing a new vaccine against TB

Tuberculosis is the world’s leading cause of death by infectious cause, and thus improving the effectiveness of vaccination is of obvious importance. The only vaccine approved for clinical use against tuberculosis, the BCG vaccine, is effective in immunizing uninfected children against M. tuberculosis infection, but of markedly variable effectiveness among adults. In a phase 2b trial, supported, conducted, and written by Glaxo Smith Kline researchers, a new vaccine against TB (a fusion of two antigenic proteins combined with an adjuvant used in other vaccines) was tested for efficacy in preventing pulmonary tuberculosis among adults already infected with M. tuberculosis. The trial took place in Kenya, South Africa, and Zambia, three African countries in which TB is endemic.

The vaccine effectiveness was 54%, comparing favorably with effectiveness of other commonly used vaccines. Serious adverse effects were infrequent. Obvious questions remain regarding whether the vaccine shows promise in reducing mortality; whether its effectiveness is comparable in other TB-endemic regions (e.g. India or Indonesia); and whether development of a new vaccine is preferable to improving the effectiveness and reliability of the already tested BCG vaccine.

Antibiotic therapy for uncomplicated acute appendicitis

What do you do with an inflamed appendix? Surgeons used to have one answer: if it’s a clinically verified appendicitis, it should be removed. However, a number of trials in recent years have begun to change minds. The APPAC trial, randomized patients in Finland aged 18 to 60 with uncomplicated appendicitis to open appendectomy or antibiotic therapy. At one year of follow up, approximately a quarter of patients suffered recurrent appendicitis.

This new study follows up the patients from the original trial at 5 years: the likelihood of late recurrence within that time was 39%, and they were not likely to have complications related to delay in surgery. This trial might not yet reflect the extent of the possible medical reversals in this area: at least one randomized control trial cited in the current study compares antibiotic-treated appendicitis to appendicitis treated with watchful waiting. As long as we have known about appendicitis, it turns out, we have also known that it might resolve on its own. Which patients can be selected for antibiotic treatment versus watchful waiting, and which are best served by surgery, remain open questions.

Hacking and healthcare data breaches

The infamously burdensome HIPAA (Health Insurance Portability and Accountability Act) was meant to provide protection for the confidentiality and privacy of patient’s health information, as was the follow up US legislation with a similarly euphonious acronym—HITECH (Health Information Technology for Economic and Clinical Health). 2149 breaches of healthcare data have been reported to the US Department of Health and Human Services, comprising 176.4 million records. Researchers classified these breaches into category (from what entity the data came), media location (e.g. desktop, email, laptop, paper record), and type. The shift in breach type was the most notable finding in this study: as records have shifted from paper to electronic formats, hacking has become by far the most common type of breach.

The authors conclude that networked health information has the potential to harm patients through data breaches. But the most salient limitation of their study, to my mind, is one they do not mention at all: a significant proportion of health-relevant information (including information about substance use and mental health) is now found in social media and under the control of under regulated, monopolistic social media concerns. If electoral breaches in the UK and the US have been found in the purview of social media platforms, is health information any safer? How do we protect against health information breaches when social networks are unsafe?

Effect of anticoagulation on stroke rates

This is probably the most important study that will be published this year regarding the effect of anticoagulation on stroke rates—but it’s not a trial. It’s a simulation, using a Markov state decision-analytic model, of the benefit due to anticoagulation given differences in assumed rates of stroke, which can vary by an order of magnitude between trials. Guidelines use a risk index like the CHA2DS2-VASc to estimate the risk of stroke in a patient with atrial fibrillation, recommending for or against anti-coagulation at a given cut-off of that index. But there’s one problem: there is significant heterogeneity in stroke rates in non-anticoagulated patients with atrial fibrillation which is smoothed out by those risk estimates.

The authors therefore simulated the potential variability of benefit due to anticoagulation, based on that heterogeneity in actual stroke risk. Here there’s another complication: they used quality-adjusted life years to estimate such benefits. This is a controversial technique which assumes patients’ experience of stroke is broadly quantifiable and generalizable. In any case, the major point is that cut-offs of these risk scores do not provide clear answers to anticoagulation recommendations

What is the best way to proceed? I’m going to stick my neck out and disagree with the authors. As useful as more precise stroke estimates (their recommendation to improve matters) might be a serious examination and clarification of patient preferences. Doctors can be trained to do that better. Even if we revise the CHA2DS2-VASc, our goal should not be to box patients into the recommendation, but to empower their own preferences.

The burden of disease in Russia

The massive, productive Global Burden of Disease study has been previously discussed in these reviews. GBD researchers compared morbidity and mortality in Russia from 1980 to 2016, attributed to 333 causes, and modified by 84 risk factors, to GBD estimates in countries neighboring Russia, and to comparable high-income countries. During this time span, life expectancy increased, but the gap in life expectancy between men and women widened (Russian men have lower life expectancy). While the decreases in life expectancy during the social and political upheavals of the 1990s seem to have been recouped, major challenges continue, including the contribution of alcohol to many attributable causes of morbidity and mortality. The large number of attributable comparisons, as well as the heterogeneity of large geopolitical entities like Russia (not to mention the fact that morbidity and mortality of certain groups, e.g. LGBTQ, is likely underreported in the current political environment), make it difficult to say how reliable these time trends might be.

Association between physician USNWR medical school ranking and patient outcomes and costs of care

Our employers, colleagues, and patients pay close attention to the US News & World Report (USNWR) rankings of hospitals and medical schools. But do they mean anything with regard to quality? More specifically, are the USNWR rankings of a given medical school associated with mortality, readmission, or cost of patients admitted to affiliated hospitals? Hospital quality depends on a number of factors, including referral and payment systems, organizational characteristics, and processes of care—not to mention patient experience of care. Medical school ranking is a poor proxy for all of these; thus, the lack of association found in this study is not surprising. This is not to criticize the work of these researchers, who executed a number of statistical techniques and careful controls to make sure that their analysis accounted for the nuances of physician attribution and patient risk. They also considered whether alternative rankings, e.g. “social-mission-centric medicine,” or NIH funding, were associated with quality; they were not.

These results should not be overgeneralized. We know that hospitals differ, and that practice differences affect quality (patient volume, specialist experience, and teaching versus community status are all attested as relevant differences). That USNWR rankings are not associated with mortality means that the search for useful, generalizable predictors is still on.

Evaluation of the causal effects between subjective wellbeing and cardiometabolic health

I lack the expertise to evaluate the methods of a Mendelian randomization study such as this. I will state the assumptions underlying the methods, however, and talk about what we can learn from the conclusions. A Mendelian randomization study uses genomic data as an instrumental variable, correlated with an independent variable and thus with the potential to clarify that relationship between it and a dependent variable. Like any statistical technique, Mendelian randomization is dependent on assumptions, including that there is no causal pathway linking the genotype and any potential confounders.

Thus, any conclusion such as one of those reached in the study, “we have found evidence indicating that higher body mass index has a causal relation with lower subjective well being,” should keep potential limitations in view. I did not see attention to potential confounders. Social determinants of health, and the recognition that there is more on heaven and earth than the physiology of the individual, their behaviour, and their diagnoses, does not seem to have a place here. To take one example from my daily clinical environment (albeit one affecting a different population than this study’s): Baltimore, Maryland, has a dearth of green space (its “urban tree canopy” is markedly less than other cities’), and notable food deserts—both due to deliberate policies privileging richer, whiter areas of the city over areas where African-Americans predominate. Such policies also affect well-being of Baltimoreans. I imagine that these social and economic forces might also be relevant in the UK. That is, social inequities of long standing can help cause both obesity and low well being; and these are more modifiable than the genome.

Sleep disturbance in hospitalised patients

Have you ever been hospitalized? Was your sleep lousy? What about taking care of someone in hospital—did that patient get a good night’s sleep? We have plenty of anecdotal experiences, but what factors are associated with poor sleep in hospital, and is there a quantitative difference between sleep in hospital and at home? Empirical research to the rescue! You can imagine half-a-dozen ways to do this sort of study, but my sleeping cap is off to the Dutch researchers whose “flash mob” research included about 2000 patients in 39 hospitals on a single day—22 February 2017. (Clearly, these results are not generalizable to all patients, as those who were too ill or unable to consent were not included.)

Patients reported they slept about 83 minutes less in hospital than at home, mostly attributable to being woken earlier. They were also woken more often during the night (about 3 times on average in the hospital versus 2 times at home). What factors did they attribute to poor sleep? Noisy neighbors, being woken by staff, and getting up to use the bathroom.

These findings are not earth-shaking but neither are they soporific. They are the small details that can matter a great deal to individual experiences and significantly affect health. Now let us go forward and let our patients sleep unless it’s absolutely necessary to wake them—and, what’s more, make policies that encourage sleep as a matter of course.

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