Healthcare rationing around the world

Approaches to rationing in pandemic and non-pandemic scenarios

Akshay Swaminathan
7 min readMar 31, 2021
Photo by Christina Victoria Craft on Unsplash

The NEJM recently published guidelines for healthcare rationing during COVID-19 that center on four values:

1. Maximizing benefit: Save as many lives or life-years as possible. If patient A will survive 3 years after treatment and patient B will survive 5 years after treatment, treat patient B.

2. Promoting and rewarding instrumental value: Prioritize those who saved or will save lives. If patient A is a doctor and patient B is a legal aid, prioritize patient A to receive a ventilator.

3. Treating people equally: If patient A and patient B have similar prognosis, randomly decide who receives treatment first.

4. Giving priority to the worst off: “Worst off” could be defined in many ways, but they define it as “sickest” (worst in terms of health) and “youngest” (worst in terms of having lived the shortest lives). If patient A is 5 years old and patient B is 50 years old and they have similar prognoses, prioritize patient A.

These values are not perfect

For example:

  • No explicit considerations for bias and discrimination. Applying these values relies on objective assessment of patient prognosis, but we know that needs of Black patients are more likely to be underestimated by physicians.
  • Prioritizing younger patients ignores dependents of older patients. Although treating a younger patient may save more life-years, this approach ignores information about dependents. It’s unclear whether treating a 5-year old with no dependents or treating a 50-year old with four dependents is the best option in terms of maximizing life-years saved. The authors argue that it’s difficult to obtain information about dependents during a pandemic and that incorporating such information risks making judgments about social worth. One alternative could be to weight life-years towards middle age, when people are most likely to have dependents.
  • No considerations for distributive justice. We know that certain groups, like incarcerated individuals, have less access to medical care than the general population (can’t we at least give prisoners soap?). Should we account for these baseline inequities when deciding how to ration care?
  • Rewarding participation in clinical trials is problematic. The authors suggest that prior participation in research to prove safety and efficacy of vaccines and treatments should be used as a tiebreaker when deciding to treat two otherwise equivalent patients. Opportunities to participate in clinical trials are not equally accessible to everyone, especially minorities and patients with certain conditions (this study found that 35% of COVID-19 clinical studies excluded pregnant women, 10% excluded those with cancer, and 16% excluded women all together).

Applying these values in South Africa, Zambia, India, and the UK

Although the framework described above were intended for pandemic scenarios, it’s interesting to apply the same reasoning to diverse instances of health care rationing.

In 2010, the World covered health care rationing in several countries. Using the information provided in this series of articles, let’s evaluate the rationing approach in these countries using the four values laid out above.

South Africa: Rationing by committee

Problem: Dialysis is expensive in South Africa ($10k per person per year), so certain health centers ration dialysis to save resources for other forms of treatment.

Method: A group of a dozen healthcare professionals would decide which patients would receive dialysis. The committee consisted of the attending nephrologists, responsible physician, social worker, renal nursing staff, and renal technologists. Decisions were often made by estimating the potential contribution of patients to society. About 1 in 5 patients was accepted for treatment.

Maximizing benefit: While a patient’s potential benefit from dialysis was considered, maximizing life-years saved was not the primary goal of deliberations, as indicated by the committee’s consideration of patients’ “social worth”.

Promote instrumental value: In the NEJM article, instrumental value refers specifically to frontline healthcare workers. This strict definition makes sense in a pandemic where frontline healthcare workers play a key role in managing outbreaks. In a non-pandemic scenario, it’s less clear whether instrumental value of certain lives should be considered and if so, how it should be defined. Nevertheless, the committee considered factors like employment, marital status, dependents, substance use, and lifestyle when determining a patient’s instrumental value or social worth. With such a complex set of criteria, it’s difficult to measure instrumental value objectively without introducing bias and discrimination.

Treating people equally: There is no indication that patients with similar prognosis were treated equally. Two physicians from a health center that performed rationing by committee in South Africa detailed the decisions of their committee over 15 years. They found that the patients mostly likely to be accepted for renal replacement therapy were aged 20–40 years, white, employed, married, non-diabetic, and lived in proximity to a dialysis center.

Priority to the worst off: Not considered explicitly.

UK: Rationing by cost

Problem: In the UK, the government provides and pays for health care through the National Health Service (NHS). In order to control spending, the NHS only reimburses medical treatments that are deemed “cost-effective”.

Method: The UK National Institute for Health and Clinical Excellence (NICE) conducts cost-effectiveness analyses to determine which treatments should be reimbursed by the NHS. Treatments that exceed NICE’s cost-effectiveness threshold (~$45,000 per quality-adjusted life year gained) are usually not reimbursed and must be paid for out of pocket.

Maximizing benefit: By capping the cost of a life-year, NICE ensures that resources are saved for the most cost-effective treatments. However, this system does not maximize benefit per dollar spent, rather it increases on average the benefit per dollar spent. In the case of dialysis rationing in South Africa or ventilator allocation in the US, rationing decisions are based on individual patient prognoses (“How long would this specific patient survive after treatment?”). In the UK, drug rationing decisions are not based on individual patient prognoses but rather population-level aggregate measures of prognosis (“How long would the average patient survive after treatment?”). This short-changes patients whose prognoses are above average.

For example, NICE ruled that bevacizumab, a colorectal cancer drug, did not meet the cost-effectiveness threshold for reimbursement because it extended life by an average of six weeks (based on clinical trials) and was priced at ~$30,000 per person per year. This decision restricts access to bevacizumab for all patients, even those whose prognoses may be greater than six weeks.

In an ideal (imaginary for now) world, reimbursement would occur at the individual level: a doctor prescribes treatment X to John, and the payer (NHS in the UK) calculates how much John would benefit from treatment X and reimburses accordingly.

Promote instrumental value: NICE does not consider instrumental value of patients in recommending drugs for reimbursement. All NHS patients are subject to the same reimbursement rules. In a non-pandemic scenario, this is sensible.

Treating people equally: The UK’s rationing decisions are individual-agnostic. If a drug is reimbursed by the NHS, everyone has access to it, whether their prognosis is good or bad. The flip side: if a drug is not reimbursed, then only those who can afford to pay out of pocket have access.

Priority to the worst off: Because the UK’s rationing decisions are individual-agnostic there is no mechanism to give priority to the worst off. A side effect of this system is that treatments for rare illnesses may be less likely to be reimbursed because such treatments are often priced higher given the smaller demand.

Zambia: Rationing by queue

Problem: In Zambia, antiretrovirals (ARVs) to treat HIV are free, but distribution systems to ensure everyone gets the medication they need are not robust. About 300,000 people across the country take ARVs.

Method: Although there is no explicit rationing plan in Zambia, the current distribution system requires patients to wait in long lines for hours before being seen by a professional and getting the ARVs. This imposes “implicit rationing,” whereby those who manage to line up first are prioritized. No explicit decisions regarding priorities, values, or criteria are made, and no conversations regarding rationing take place. It’s a way to ration without having to deal with the ethical implications of explicit rationing.

Patients may wait up to seven hours, which is prohibitively long for those who need to work to support their families. These issues lead to noncompliance, where patients stop picking up their ARVs or showing up for appointments because they cannot or do not want to wait on line.

Maximizing benefit: Not considered.

Promote instrumental value: Not considered.

Treating people equally: Not considered. Those who live near clinics and those who can afford to wait in long lines are implicitly prioritized.

Priority to the worst off: Not considered.

India: Rationing during H1N1

When deciding how to allocate ventilators during swine flu, patients with the best prognosis are given priority. At the same time, policy forbids removing patients off ventilators. Managed to create makeshift ventilators. Essentially they got lucky.

Problem: In Sassoon General hospital in Pune, India, the demand for ventilators exceeded the supply during the 2009 H1N1 pandemic.

Method: Patients with the best prognosis were prioritized, but hospital policy forbade removing ventilators from patients with poor prognosis. One doctor designed a makeshift ventilator that helped satisfy demand when all automatic ventilators were occupied.

Maximizing benefit: The hospital tried to maximize benefit by prioritizing patients with the best prognosis. At the same time, the policy of not removing ventilators from patients to make room for those with better prognosis contradicts the “maximizing benefit” principle. The NEJM article states that removing patients from ventilators without consent is justifiable, though the authors acknowledge it may be traumatic for physicians and patients.

In a response letter to the NEJM piece, Liddell and colleagues state that removing patients from ventilators is considered a crime in the UK and other European countries. In another letter, Firth and Eyal argue that “Withdrawal of care from a patient may also harm staff morale and public trust, which are essential resources during times of emergency.”

Responding to these critiques, Emanuel and colleagues call prohibiting reallocation a Pyrrhic victory — “it would avoid the tragedy of ventilator removal only by proliferating tragic situations in which people with good prognoses die for want of a ventilator.”

Promote instrumental value: Not considered.

Treating people equally: The article doesn’t describe how patients with equal prognosis would be allocated ventilators.

Priority to the worst off: Not considered.



Akshay Swaminathan

Data scientist, global health researcher, language learner