Which factors should be included in triage? An online survey of the attitudes of the UK general public to pandemic triage dilemmas

Abstract

Objective: As cases of COVID-19 infections surge, concerns have renewed about intensive care units (ICUs) being overwhelmed and the need for specific triage protocols over winter. This study aimed to help inform triage guidance by exploring the views of lay people about factors to include in triage decisions.

Design, setting and participants: Online survey between 29th of May and 22nd of June 2020 based on hypothetical triage dilemmas. Participants recruited from existing market research panels, representative of the UK general population. Scenarios were presented in which a single ventilator is available, and two patients require ICU admission and ventilation. Patients differed in one of: chance of survival, life expectancy, age, expected length of treatment, disability and degree of frailty. Respondents were given the option of choosing one patient to treat or tossing a coin to decide.

Results: Seven hundred and sixty-three participated. A majority of respondents prioritised patients who would have a higher chance of survival (72%-93%), longer life expectancy (78%-83%), required shorter duration of treatment (88%-94%), were younger (71%-79%) or had a lesser degree of frailty (60%-69%, all p<0.001). Where there was a small difference between two patients, a larger proportion elected to toss a coin to decide which patient to treat. A majority (58%-86%) were prepared to withdraw treatment from a patient in intensive care who had a lower chance of survival than another patient currently presenting with COVID-19. Respondents also indicated a willingness to give higher priority to healthcare workers and to patients with young children.

Conclusion: Members of the UK general public potentially support a broadly utilitarian approach to ICU triage in the face of overwhelming need. Survey respondents endorsed the relevance of patient factors currently included in triage guidance, but also factors not currently included. They supported the permissibility of reallocating treatment in a pandemic.

Keywords: covid-19; intensive & critical care; medical ethics.

Figure 1

Hypothetical pandemic triage dilemmas. (A) Example question with varying chance of survival; (B) example question with varying age; (C) example question with varying degrees of disability. ICU, intensive care unit.

Figure 2

(A) Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different survival chances. There was a statistically different distribution in responses when comparing the 80% vs 10% with the 40% vs 10% chances of survival scenario (Χ² (3, N=763)=19.793, p<0.001), 80% vs 10% with the 20% vs 10% scenario (Χ² (3, N=763)=165.077, p<0.001), 80% vs 10% with the 51% vs 49% chances of survival scenario (Χ² (3, N=763)=371.54, p<0.001). Similarly, we found statistically different distribution in responses when comparing the 40% vs 10% with the 20% vs 10% scenario (Χ² (3, N=763)=143.00, p<0.001); 51% vs 49% scenario (Χ² (3, N=763)=351.298, p<0.001) and 20% vs 10% with the 51% vs 49% (Χ² (3, N=763)=198.278, p<0.001). (B) Respondent choices in a triage dilemma involving patients with different survival chances where the patient with worse prognosis was already receiving treatment in intensive care. There was a statistically different distribution in responses when comparing these scenarios with equivalent withholding versions: 80/10: Χ² (3, N=763)=27.766, p<0.001; 2: 40/10: Χ² (3, N=763)=87.105, p<0.001; 3: 20/10: Χ² (3, N=763)=81.977, p<0.001; 51/49 Χ² (3, N=763)=180.061, p<0.001.

Figure 3

Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different life expectancy. There was a statistically different distribution in responses: 1: 25/5 years vs 40/15 years: Χ² (3, N=763)=23.89, p<0.001; 2: 25/5 vs 25/15: Χ² (3, N=763)=62.562, p<0.001; 3: 25/5 vs 15/14: Χ² (3, N=763)=305.042, p<0.001.

Figure 4

(A) Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different age but identical survival chance/life expectancy. There was a statistically different distribution in responses: 1: 82/55 vs 82/66: Χ² (3, N=763)=19.455, p<0.001; 2: 82/55 vs 71/55: Χ² (3, N=763)=47.608, p<0.001; 3: 82/66 vs 71/55: Χ² (3, N=763)=25.64, p<0.001. Responses on all scenarios differed significantly when compared with the 72/71 scenario, Χ²>1061.42, p<0.001. (B) Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different age where the older patient had a lower survival chance. There was a significant difference in distribution of responses compared with equivalent scenarios where survival chance was said to be identical: 71/55: −Χ² (3, N=763)=95.65, p<0.001; 82/55: Χ² (3, N=763)=77.219, p<0.001.

Figure 5

Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different expected duration of treatment. There was a significant difference in the distribution of answers between scenarios: 1: 24 weeks/1 week vs 1 week/2 weeks: Χ² (3, N=763)=42.66, p<0.001; 2: 24 weeks/1 week vs 5 days/1 day: Χ² (3, N=763)=30.047, p<0.001; 3: 24 weeks/1 week vs 10 weeks/1 week: Χ² (3, N=763)=1.085, p=0.99; 4: 2 weeks/1 week vs 5 days/1 day: Χ² (3, N=763)=2.798, p=0.99. ICU, intensive care unit.

Figure 6

Respondent choices in a triage dilemma involving withholding treatment from one of two patients with different degrees of pre-existing disability. There was a significant difference in the distribution of answers between scenarios: 1: profound learning disability (LD)/none vs moderate/mild Χ² (2, N=763)=536.177, p<0.0001; 2: profound LD/none vs physical/none Χ² (2, N=763)=464.653, p<0.0001.