A carer in a nursing home asked this question on seeing Mrs J, an 82 year old lady with Alzheimer’s disease, push away food and clamp her mouth shut when offered a drink. Mrs J had been deteriorating for many months. Thin, bed bound, incontinent, unable to speak except to grimace or occasionally shriek, she needed help with all activities of life, able to do nothing for herself. Her appetite, never huge, had gone entirely. Now she was refusing everything, pushing carers away when trying to feed her, turning her head away when drinks were brought to her mouth. The staff were having a “What Do You Do When You Don’t Know What To Do” moment, looking to me for an answer; I felt I was having a WDYDWYDKWTD moment of my own.
Effective Palliative care can be tricky; knowing when palliation turns into end of life care, even harder. Could a structured approach to uncertainty help me here?
Mrs J’s care involved both diagnosis and management issues. The types of uncertainty here were ANALYSING (what were the diagnoses?) and TEAM-WORKING (arranging the right care with a group team or family) because Joan lacked capacity to plan or negotiate care herself, except perhaps, by refusing sustenance.
Using ANALYZING skills from the Map of Uncertainty in Medicine I first hypothesised and then ruled out intercurrent illness. Apart from the feeding issues nothing else had changed, examination was unremarkable. Although a little dehydrated she did not appear to have an infection, or be in pain, or have any mouth issues. Her dementia had been relentlessly progressive, with less of her old self apparent every week. Even offering her favourite foods did not help. Thus, I was able to rule in a diagnosis; this was a development of her disease, a sign that we were reaching the end stage of dementia, when even food is no longer recognised.
Then I turned to the TEAM-WORKING quadrant. This was not just about me and Mrs J. Decisions involved her carers and nurses, and her two sons, who lived nearby. There are different ways to plan care. Pathways are for straightforward and slow moving conditions (a hernia repair in a fit 40 year old). Algorithms help when things move fast but are single conditions (childbirth), whereas pattern recognition helps more when things are fast moving and complex ( ITU, major trauma). This was a slower moving, complex situation which is also strongly preference sensitive, meaning care needs to be individualised to the person. Scenario planning is what’s needed here.
Fortunately, when Mrs J was admitted we had met with her family to do just that, and staff had regular updates with her sons as things deteriorated, creating an Advance Care Plan (1). Although saddened, they recognised that Alzheimer’s was taking their mother inexorably away from them. They wanted Mrs J to stay in situ; her last hospital admission had made her fearful, distressed and isolated on a busy ward. They also wanted to “Allow a Natural Death” (AND), a kinder and more accurate phrase than “Do Not attempt Resuscitation”. We had anticipated various scenarios including this one; how eventually the dementia could even result in her failing to recognise food and drink.
We called them once more, sharing the perception that Joan was approaching the end of her life; they were relieved to have this in the open having thought as much themselves. She was nursed tenderly in her own bed with her familiar carers and her sons visiting often. She slipped into a coma and died peacefully a few days later.
And no, we didn’t let her “starve to death”. She died from her Alzheimer’s disease. Emaciated patients with cancer do not starve to death, but die from their disease too. Working with the whole team, recognising her family as a crucial part of that team, meant that, for Mrs J, the uncertainty generated by the pursed lips and refusal of drinks, resulted in careful planning and a peaceful death.
Her sons sent the staff a big bunch of flowers and thanked them for their care.