How the risk:benefit of bisoprolol became a medic:psychiatry trade off

A patient has anoxic brain injury followed an MI a few months ago. He was fit and well both mentally and physically before this, with only well controlled asthma blotting his otherwise pristine medical record. The brain injury has left him with mostly frontal signs. He has been disinhibited and had been becoming increasingly aggressive. Several members of the nursing staff had been at risk from physical aggression.

Post MI, he was started on the usual medications:

  • Simvastatin 40mg NOCTE
  • Bisoprolol 1.25mg OD (down from 1.25mg BD a few weeks before)
  • Ramipril 2.5mg BD
  • Clopidogrel 75mg OD
  • Aspirin 75mg OD

For the psychiatric symptoms, he was been on various atypical antipsychotics. The concern with most of them is the QT prolongation, especially post MI. Aripiprazole seems to have the least effect on the QT interval and was initially used. However, this failed to adequately control his symptoms. Quetiapine seemed to be more effective, but came with a risk of QTc prolongation.

He was also on diazepam 5mg TDS, with an aim to titrate down if his behaviour allowed. He also had PRN diazepam and quetiapine.

Every morning, his observations were relatively consistent. His pulse was between 50 and 60 and his blood pressure about 95-105/60. His saturations would vary considerably, dropping as low a 84%. They would pick up with a salbutamol nebuliser. He would have about 2-3 episodes of sats in the low 90s a week. Since his anoxic brain injury, he has not been compliant with inhaled medication, even with a spacer.

I was called to authorise PRN diazepam when he was being particularly aggressive with staff whilst having saturations of 92%, BP 94/47 and pulse 48. This was a difficult call. I could hear a wheeze in between his verbal threats, so I decided to give a salbutamol nebuliser first, which I hoped would perhaps pick up his pulse as well as improve his saturations. He tolerated it remarkably well, although his saturations stayed the same afterwards, with a pulse of 51 and BP 104/67. I decided to give 2mg of diazepam and observe closely.

Diazepam given orally has a rapid onset of action, and takes 1 hour to reach its peak effect. It then has a half life in the region of 20-50 hours.  Once he has got through the first hour, it is less likely he will then go into respiratory depression/cardiovascular collapse, which is why particularly close attention to his obs is paid in that crucial first hour.

This case illustrated the difficulty of managing antipsychotics and benzodiazepines in a patient demonstrating aggression with a previous MI and low pulse, BP and saturations. The aggression has been getting quite serious; staff have been injured. How could we tweak the medications to get the most benefit medically and also boost his BP, pulse and asthma control to make PRN doses of diazepam safer?

Plan A: Stop bisoprolol

In favour: This would increase his pulse and blood pressure, and we could always titrate ramipril for any worsening blood pressure control. It may also help his asthma.With improved observations, we could give diazepam more safely as needed.

Against: How much benefit from post MI beta blockade would we be sacrificing? And is it even likely that such a small dose of a cardioselective beta blocker was really worsening his asthma?

I discussed this with a medical consultant. He said that it was unlikely that bisoprolol would be to blame for the asthma worsening, and it was far more likely that lack of compliance with his medications was the problem. He said that if the flares continued, it would be worth stopping bisoprolol and titrating ramipril.

I still wanted to know how much benefit it would cost to drop the beta blocker. There was a summary published in Circulation in 2002. It was first shown in 1965 that giving propanolol indefinitely post MI improved mortality. A RCT in 1982 was PICO’d as follows:

P: 3,837 post MI patients aged 30-69
I: Daily propanolol (180/240mg OD, depending on serum levels) started 5-21 days post MI, continued over the follow up period (average 24 months)
C: Placebo
O: Total mortality during the average 24-month follow-up period was 7.2% in the propranolol group and 9.8% in the placebo group

With the addition of ACE-inhibitors, PCI, antiplatelets and all other exciting treatments since this trial, some then questioned whether beta blockers really have that much benefit compared to the new treatments, and whether the absolute benefit from the beta blockers was reduced now that there were so many other interventions to reduce mortality. The SAVE trial looked at whether beta blockers made a difference post MI over and above ACE inhibitors. They did.

The next question was whether cardioselective beta blockers are safe in people with asthma. The only meta-analysis I found looking at this question directly found that in patients with mild to moderate airway reactivity, the first dose of a cardioselective beta blocker does reduce FEV1, but over continous dosing this effect diminishes. This meta-analysis had asthma patients with a mean age of 40 however, which is much younger than the age of most patients on beta blockers. It also did not look at patients with severe asthma.

Comparing the risk of asthma against the risk of MI alone, bisoprolol should probably continue.

The complicating factor was the observations during the day at times when the patient was aggressive and needed PRN diazepam. The patient was clearly agitated at times and improved considerably on diazepam. Would it not be better for his mental health and safer for the staff to be able to safely use diazepam when all else fails? You could argue we should treat the patient not the obs, and a patient who is up and about and swinging violently clearly has a more than decent A, B and C. My concern is that whilst at the time of considering diazepam he seemed well, if he were to have a bad reaction then with a PMHx of unmedicated asthma and previous catastrophic MI, the risks of respiratory depression and cardiovascular collapse are probably heightened. It is in his medical interest to ensure his observations were optimised to give him more slack to comfortably tolerate diazepam doses.

Plan B: Nebulised steroid daily as preventative

The patient was able to tolerate nebulised salbutamol, even though he would not use a spacer. Perhaps it would be worth using step 2 of the BTS ladder, giving the same dose of steroid (200 micrograms BD) via a nebuliser to help his long term control.

It would be great to hear what, if any, changes to his medications you would make. What do you think should be done?


Hyperglycemia as a sign of something systemically wrong

The design of the new psychiatry unit was so damn mathematical. Everything was coloured in single hues and with the minimum number of edges possible. It was as if someone had designed it on an original PSOne and needed to keep the textures and polygon count simple, and was sponsored by a consortium of Euclid, Pythagorus and MS Paint.

It was a welcome change from the surgical attachment. Towards the end of the surgical placement we needed cover for the twilight period from pm to midnight, as the doctor down for that shift had called in sick. I decided to go for it. I was getting tired of all things surgical but thought that on my next job I might come to miss all this. How wrong I was.

At 6pm I was called to see a man in his late 40s day 5 post abdominal surgery. I was asked to review his capillary glucose around the 20-23 mark. He had a background of Type 2 diabetes.

The first thing I asked for was his ketones, which came back as 0.1. This excluded DKA for now.

My next priority was to consider the possibility of hyperosmolar hypgerglycemia state (HHS), previously known as HONK. I checked the Diabetes UK Guidance, which stated three criteria for HHS:

  1. Hypovolemia
  2. Glucose > 30mmol/L
  3. Hyperosmolar (>320 mosm/kg)

So now I knew my patient was not in DKA, and not in HHS. I could relax just a little. However, he could still be dehydrated from hyperglycemia, and so a fluid status assessment was needed. I also needed to look for an underlying cause.

I see fluid status assessment as a 3 part process:

  1. How is the patient clinically
  2. How are the patient’s observations
  3. How is the patient biochemically

If you feel the patient is shocked, you want to be able to work out which type of shock is most likely.

The aim of the game is to keep spongey Pacman hydrated. If you lack fluid in the tank, lack a decent pump or can't get enough pressure at the end, Pacman will dry up. B
The aim of the game is to keep spongey Pacman hydrated. If you lack fluid in the tank, lack a decent pump or can’t get enough pressure at the end, Pacman will dry up. This corresponds to the main three types of shock.

It’s usually possible to work out the type of shock (hypovolemic, cardiogenic or distributive) from the  history, clinical status, the obs and the biochemistry.

(For completeness, there’s also obstructive shock in say massive PE, and hypoadrenal shock as in an Addisonian crisis).

I will go through fluid balance in more detail in a future post. In this patient’s case, there was an increased pulse, and the rest of his observations were normal. He was not catheterised, but was passing some urine every few hours. He complained of no new pain, and in particular no abdominal pain (anastomotic leak is the concern here at this time in the postoperative period). His mucous membranes seemed a little dry.

His chest was clear and a urine dipstick was normal. He had no lines.

I looked at his blood results from that morning.

3/4/13 1/4/13
White cell count 20.2 8.3
eGFR 33 >90

There were no medications to explain the AKI, or any recent intravenous contrast. He had no features of obstruction. The most likely cause was pre-renal. This is problem  #1.

Problem #2 is the glucose. Although not a medical emergency in itself yet, this was an acute rise and needed to be explained. He had taken his medications as normal, and had reasonably well controlled glucose before admission. His post operative recovery until then was plain sailing, and his post op glucose had never been above 13 until today.

In addition to medications, the four causes we must always consider for sudden hyperglycaemia in a diabetic are ACS, stroke, pancreatitis and infection.

In terms of ACS, he had no chest pain. However, we must be cautious about the lack of chest pain in diabetics, the elderly and females. In fact, one third of all ACS cases have no chest pain. In addition, a patient over the age of 85 who is having an MI is more likely to complain about shortness of breath than chest pain. Have a listen to the amazing Canadian Emergency Medicine Cases podcasts, free to F1s and medical students. In this case the ECG showed no changes and initial and 12 hour troponins were negative.

I then looked at the rest of his obs, and mentally overlaid the glucose chart with the obs chart. This is what I saw:

Obs chart

Hmm. Was there a unifying diagnosis to explain the AKI, the hyperglycemia, the pulse and the acutely elevated WCC in a patient with no fever, no pain and no localising signs of infection?

This patient met SIRS criteria (pulse, WCC). Even though there were no localising signs, it was still clinically justified to consider the abdomen as a source of sepsis given the history (SIRS + source = sepsis). In addition, there was end-organ damage (AKI). I felt that this patient therefore had severe sepsis.

It was a handy coincidence that just the day before I had written a question for my iPhone app about osteomyelitis in a diabetic patient. I was looking up information about how osteomyelitis presents in diabetic patients compared to the classic textbook description, and found the following:

“Elevated serum glucose can be a marker of systemic and less often local infection in the diabetic patient. In their history, the compliant patient will often complain of long-standing difficulty with elevated blood sugars roughly corresponding with the time frame of the presence of their ulcer. Often, recalcitrant hyperglycemia is the only systemic manifestation of osteomyelitis in these patients.” –

I read around some more, and found that sepsis and hyperglycemia are well associated. It’s probably part of the acute stress response.  Whether or not treating the hyperglycemia improves outcomes in sepsis seems to be less clear.

Edit – 10/4/13 The patient turned out to eventually have an ischaemic bowel after all that. This would explain the raised lactate, and the SIRS criteria. In ischaemic bowel, the physical findings are classically out of proportion to the degree of pain. In the early stages, there may be minimal or no tenderness and no signs of peritonitis. In the later stages typical symptoms of peritonism develop, with rebound guarding and tenderness.