Diabetes & Chronic Disease Management in Advanced Illness

The fundamental shift in goals

As prognosis shortens, diabetes management pivots from preventing long-term complications to symptom control and quality of life. The right approach depends on disease trajectory and expected prognosis, not a fixed target.

Target glucose ranges by prognosis

Prognosis	Target range	Approach
Weeks–months	6–15 mmol/L	Continue monitoring; consider reducing medication
Days	<20 mmol/L	Reduce/cease monitoring frequency
Hours–days (end-of-life)	Symptom-based only	Cease all monitoring and glucose-lowering drugs

Avoid HbA1c testing in this population — results are unreliable (transfusions, liver disease, renal failure all distort it) and don’t align with the actual goals of care at this stage. Continuous glucose monitors become inaccurate at end-of-life due to poor tissue perfusion, and their readings can distress caregivers without offering any clinical benefit. Relax glycaemic targets significantly — asymptomatic hyperglycaemia is acceptable as disease progresses.

Why targets relax: the evidence base

The rationale for relaxing targets isn’t just pragmatism — it’s grounded in the two landmark cardiovascular-outcome trials that shaped modern glycaemic-target guidance. ACCORD randomised over 10,000 people with poorly controlled, long-standing type 2 diabetes to an intensive glucose-lowering arm (HbA1c under 6%) or a standard arm (HbA1c 7–7.9%). The ACCORD study showed a 22% relative increase in total mortality in the intensive glucose lowering arm, and severe hypoglycaemia requiring medical assistance was three times more common in the intensive group. Neither ACCORD nor the comparable ADVANCE study showed that intensive glucose lowering reduces macrovascular events compared to standard glucose lowering in older individuals with a long history of diabetes, and rapid, intensive glucose lowering could be harmful in this high-risk group — the same population that overlaps heavily with advanced/palliative-stage diabetes management.

Evidence specifically for de-intensifying (deprescribing) diabetes medicines is thin. Little evidence is available on the effects of discontinuing drugs for diabetes, and a systematic review identified only two low-quality controlled studies of deprescribing these drugs; in both studies, there was no significant difference in HbA1c between the group that discontinued diabetes drugs and those that continued. Tight glycaemic control may be appropriate in those who have sufficient life expectancy to benefit from the reduced risk of microvascular complications, but for frail older people intensive therapy increases the risk of hypoglycaemia without a mortality benefit — de-intensification of therapy is therefore often appropriate. General deprescribing guidance also flags long-acting sulfonylureas (alongside benzodiazepines, antipsychotics, and NSAIDs) as a class where a dose taper, not an abrupt stop, is the safer way to withdraw them.

De-intensification: what to stop first vs continue longer

Stop or reduce first (higher hypoglycaemia/harm risk in this context):

• Sulfonylureas (gliclazide, glipizide) — taper rather than stop abruptly if long-acting, per general deprescribing guidance
• Long-acting insulin (unless there’s genuine beta-cell failure)
• GLP-1 agonists (worsen anorexia — counterproductive in a population where appetite loss is already a major issue)
• SGLT2 inhibitors (ketoacidosis risk compounded by poor oral intake)

Continue longer (comparatively safer profile):

• Metformin (unless renal/hepatic dysfunction)
• DPP-4 inhibitors
• Short-acting insulin, if the patient is eating unpredictably

Important exception: patients with beta-cell burnout or type 3c diabetes (e.g. post-pancreatectomy) still need ongoing insulin to prevent diabetic ketoacidosis, right up until the final hours — don’t withdraw insulin reflexively in this group. C-peptide testing can help determine residual endogenous insulin production and guide the safety of insulin cessation decisions.

Insulin regimen simplification

Use the simplest regimen appropriate to prognosis — long-acting insulin may actually cause less day-to-day variability than short-acting insulin alone in a patient with unpredictable oral intake. If the patient is on glucocorticoids, tailor timing to the drug’s pharmacokinetics (prednisolone peaks ~8 hours post-dose; dexamethasone’s hyperglycaemic effect is more prolonged).

The broader principle

This de-intensification approach — relaxing targets, simplifying regimens, and stopping medications whose harms now outweigh their benefit — applies more broadly to other chronic disease management (e.g. hypertension, hyperlipidaemia, osteoporosis therapy) as a patient’s prognosis shortens; see also the general deprescribing framework in general-medicine.