Zackary Berger’s weekly research reviews—13 December

Zackary Berger reviews the latest research from the top medical journals

Annals of Internal Medicine

Weight loss differences after bariatric surgery: which procedure is better?

Which bariatric procedure is better? The patient centered outcomes research (PCORnet) cohort study asks, among US patients, followed up for 1, 3, and 5 years, which bariatric procedure is associated with more weight loss and fewer adverse events? The researchers looked at Roux-en-Y gastric bypass (RYGB), sleeve gastrectomy (SG), and adjustable gastric banding (AGB). 

They found that RYGB is associated with more weight loss than SG, but about twice as high a prevalence of adverse events. Differences between patient groups seems to have mattered less than which surgery was performed, but those with diabetes, older people, and African American and Hispanic groups lost less weight after these procedures. Interesting to me was the higher prevalence of Hispanics among those undergoing AGB, and that more white patients underwent RYGB compared to African American and Hispanic groups. This might be due to cost differences, however, this was not addressed in this retrospective cohort study, and (due to incredible pricing heterogeneity in the US) probably couldn’t have been.

The researchers also didn’t investigate whether patient satisfaction differed between surgery types. Nevertheless, these results are very useful to anyone considering, discussing, or recommending these risky, but useful procedures.

JAMA Internal Medicine

Patients who can’t afford their insulin use less of it

I’ve had variations of this conversation with my patients a lot. “You know”—I ease into it with the patient—“I see a lot of people who stretch out their insulin to save money. Does that happen to you?” “Yeah,” they reply, “Sometimes I skip it for the last few days or week till my paycheck comes.” It’s taken me a long time to ask that question regularly.

Researchers at a Yale primary care clinic asked 199 patients about such under- or non-use of insulin owing to financial pressures; 26% reported cost-related insulin underuse. It is not surprising that such cost-related insulin underuse is more common among Latin and African-Americans patients. Also unsurprising, but dispiriting, is the fact that more than one third of patients experiencing such underuse did not report it to their doctor. There are many reasons for this, but looming large among them is the failure of clinicians to ask. The fault is to be laid at the feet of the USA’s profiteering system which cannot ensure essential medicines. I’ll quote the authors here: “Regulators and the medical community need to intervene to ensure that insulin is affordable to patients who need it.” The free market shackles poor people with diabetes.

Trauma in the hospital, rehospitalisation, and emergency department visits

Trauma due to hospital care has been receiving increasing attention. A group of Toronto based researchers sought to determine the prevalence of hospitalisation trauma, for example, disturbance in mood, mobility, sleep, or nutrition, among inpatients in two urban hospitals. They estimated its association with rehospitalisation or visit to the emergency department after discharge. There were a number of exclusion criteria, and since patients were recruited on the day of or the day before their discharge, the number of patients included was 207, which was somewhat small.

93% of participants experienced disturbance in at least one domain, with 29.5% experiencing “high trauma” i.e. disturbance in 3 or 4 of these domains. High trauma of hospitalisation was associated with a 16% greater absolute risk of 30-day readmission or emergency department visit. This is clearly a striking finding which calls out for further examination. But before we start designing screening tools to identify patients at high risk of such trauma, or intervene to prevent its sequelae, perhaps we should also ask in what direction the causal arrow goes, and how narrowly we should draw the trauma model. In other words: are patients who are more marginalized by society, more victimized or more vulnerable, more likely to experience hospital trauma, and is that a result of a pre-existing comorbidity?


Which anticoagulants are associated with higher bleeding risk?

Non-vitamin K oral anticoagulants (NOACs) are being used more, and we need more information on their real-world association with bleeding risk. Head to head randomised control trials are unlikely to compare NOAC’s manufactured by different pharmaceutical firms. This retrospective cohort study used enrollment, medical encounter, and pharmacy data of Medicare beneficiaries receiving anticoagulants. It helps quantify and compare the risk of bleeding among different NOACs (as well as warfarin), the effect of coprescription of proton pump inhibitors on such risk, and whether a gastrointestinal haemorrhage risk score was associated with hospitalization for haemorrhage. Rivaroxaban was associated with a significantly higher risk of bleeding than dabigatran, warfarin, or apixaban, in that order. With proton pump inhibitor co-therapy, the risk of hospitalization for gastrointestinal bleeding was significantly lower for each anticoagulant.

An important finding was that the difference in the association of bleeding with the individual NOACs, and the protective effect of PPI therapy, was associated with patients’ gastrointestinal hemorrhage risk. In the patients at highest risk of bleeding, the difference between the strategy associated with the most GI bleeding (rivoraxaban without PPI) and least (apixaban with PPI) was 2 hospitalizations per 100 person years. Thus the authors suggest that gastrointestinal bleeding risk be assessed before initiation of any anticoagulant. This seems reasonable and a perfect way for information integrated into the electronic health record to help inform patient preferences and decision making in the hospital.

Medicaid expansion associated with reduced mortality

Does expanding access to health insurance prevent death? The evidence in the US is mixed. While the Affordable Care Act originally mandated expansion in 2010, the Supreme Court struck down this mandate. To date, 36 states have adopted expansion of Medicaid via their own processes. This provides an opportunity to isolate the impact of expanding Medicaid using intrastate differences. Therefore, one of the most prominent groups researching the health effects of insurance asked if Medicaid expansion affected mortality among patients with end-stage renal disease.

Patients with end-stage renal disease are covered by Medicare in the US, but they are not covered for the first four months. Thus Medicaid coverage, not just in the months preceding dialysis, but in general might protect this vulnerable population. A host of sensitivity analyses, as well as post-hoc exploratory analyses, buttress the final result based on a difference-in-difference analysis: a number needed to treat of 18 additional persons needed to gain coverage to avoid one death at one year following the initiation of analysis. For comparison, the NNT for mortality over 10 years prevented by coronary artery bypass graft surgery is around 25.

The Lancet

Fluoxetine after stroke does not improve functional outcomes

A trial in 2011 and a Cochrane systematic review both found some evidence (of varying quality) for improved functional outcomes after stroke in patients prescribed selective serotonin reuptake inhibitors. On this basis, a randomized double-blinded multi site trial in UK hospitals sought to determine if fluoxetine, 20mg, given daily to patients with recent stroke, would improve functional outcomes at six months. Patients on pharmacologic treatment for depression were excluded, as well as those who did not speak English and several other categories of comorbidities. The outcome was negative: functional outcome was no different in the SSRI versus placebo groups. While there was a decrease in depression symptoms in the fluoxetine group, there was also an increase in bone fractures.

Albeit negative, this finding is informative. Problems with adjustment, as well as depressive symptoms, can be quite frequent after a stroke. At least for this particular agent, we now have at least one piece of evidence suggesting that starting an SSRI (for those without depression previously treated with medications) should not be done automatically after stroke.  

International migration and infectious disease

People have always moved from one region to another and politically many leaders have cast aspersions on the new arrivals. In terms of health outcomes, the research varies. Some shows poorer outcomes in migrant populations. On the other hand, the “healthy migrant effect” is also well known, and makes sense: healthy people are more able to migrate. A systematic review conducted by European researchers endeavored to arrive at estimates of all-cause absolute and standardized mortality ratios of international migrants (who number more than 258 million) compared to their counterparts in the host country, as well as mortality ratios for individual ICD-10-coded causes of death. The results supported the healthy migrant hypothesis, with standardised mortality ratios less than 1 for all-cause mortality for all country-level data collected. This was with the important proviso that more than 90% of the studies included were from high-income countries. This reminds me of the story of the guy looking for his keys by the lamppost because that’s where the light was best, and one might wonder if publishing such a finding, despite its positive message regarding migrants (they are healthy), might detract attention from the situation of migrants in low- and middle-income countries.

Another important finding was that infectious diseases and “external causes” (e.g. trauma) were two causes of death in which migrants—at least in this systematic review—were not healthier than populations in the host country. Those who care for migrants already know to screen for these conditions, but this review might help support those who seek appropriate resources from host countries.


Family centered communication in inpatient pediatrics: what is the key intervention?

For years, many have known that bedside rounds involving families can lead to better communication and care. Much of the inpatient communication literature consists of small, single-site studies. Thus this multisite study of a multicomponent, sophisticated intervention to place parents and their children at the centre of hospital communication in pediatric units is an important step forward. The finding that the frequency of harmful adverse events was reduced by 38% after the intervention is certainly promising.

Any nonrandomized trial raises questions. Is the positive result merely an effect of increased attention—i.e. any step towards talking more is likely to improve matters, and the big question is which single element of the intervention should be preferentially added? Such interventions are frequently difficult to sustain. How much did this one cost? Participants (e.g. patients) were compensated minimally, with snacks—would paying patients, and paying nurses (not to mention janitors) a more respectable wage lead to even greater improvement?

Finally, and frankly, I was surprised to see that this intervention lacked an explicit emotional component. Good communication requires empathy, and inviting families to talk about the dangers and fears accompanying their children’s hospital stay must involve terrifying stories of death and rescue. Explicit training on empathy might take more time, but could it be even more helpful?

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