National questionnaire survey on what influences doctors' decisions about admission to intensive care

Introduction

In the United States and Canada intensive care accounts for 20% and 8% of inpatient hospital costs, respectively.1 Fair allocation of this scarce and expensive resource is the doctors' responsibility. Although guidelines have been developed to help doctors decide on who to admit to intensive care,2^–4 they may be difficult to put into practice.5 In particular, the process by which doctors identify patients with a “reasonable prospect of substantial recovery” warranting intensive care is not well known.

Characteristics of patients that influence admission to intensive care are age, severity of illness, and reason for admission.5 6 Availability of beds has been inconsistently associated with triage decisions.6 7 Cognitive factors may also influence decisions, including biases in doctors' processing of information,8 9 the amount of information available,10 and the underlying disease.11 More specifically, and regardless of prognosis, patients with cancer may have more do not resuscitate orders and receive less cardiopulmonary resuscitation, mechanical ventilation, and intensive care.12^–16 The relative importance of these factors for decisions on admission to intensive care is not well known. We assessed what influences doctors' decisions to admit patients to intensive care. In particular we sought to determine if there was a bias against patients with cancer.

Methods

We carried out a survey of all members of the Swiss Society of Intensive Care Medicine and of doctors who had taken the intensive care certification examination in the previous two years. An anonymous questionnaire, pretested for clarity of content, was sent out in November 2001 and then one and two months later. A prepaid envelope was included for its return.

Results

Overall, 21 of 402 eligible doctors declined to participate because they no longer worked in intensive care. Of the remaining 381 doctors, 232 (61%) returned completed questionnaires. Response rates were similar across the eight versions of the questionnaire.

The mean age of respondents was 45.2 years (table 1). Most were men, worked at public hospitals, and were routinely involved in decisions on admissions to intensive care. Many had more than one certification in a medical specialty (n = 137; 59%), the most common being intensive care medicine and anaesthesiology. Most worked in surgical (n = 199; 86%) or medical intensive care (n = 159; 69%), and 15 (7%) worked in paediatric or neonatal intensive care. The mean number of beds in each intensive care unit was 11.4 (SD 6.4). Overall, 105 respondents (45%) reported that a relative or close acquaintance had ever been admitted to intensive care.

Determinants of admission to intensive care

Among the factors influencing decisions on admission to intensive care, most doctors rated as important or very important the prognosis of the acute illness and of the underlying disease and the patients' wishes (figure). Half considered the number of available beds as important. Least influential were patients' religious beliefs, drug and alcohol misuse, psychiatric history, emotional state, and socioeconomic circumstances.

Doctors' ranking of factors when assessing patients for admission to intensive care

• Download figure
• Open in new tab
• Download powerpoint

Discussion

Specific factors influence doctors' decisions to admit patients to intensive care. Some factors, such as consideration for the patient's wishes, are desirable and agree with current recommendations. Others, for example availability of beds or a patient's personality, are ethically questionable. Moreover, we found a discrepancy between the acknowledged importance given to some factors and the importance as shown by the hypothetical clinical scenarios, suggesting that unconscious processes may be at work and lead to biased decisions.

An important finding was the absence of discrimination against patients with cancer. Medical progress in the treatment of cancer and its complications gives the hope of improved survival and better quality of life. Thus doctors' perception of the therapeutic options for malignancies may be changing. The fact that the patient with AIDS presenting with haemolytic uraemic syndrome was more often admitted to intensive care than the patient with breast cancer does not necessarily imply discrimination. Haemolytic uraemic syndrome is a rare occurrence in patients with HIV infection and has a grave prognosis. The doctors in our study may have thought that HIV infection was more amenable to treatment than advanced breast cancer. Insufficient knowledge rather than bias would account for the difference in admission rates between the two patients.

Doctors rated the potential determinants of decisions for admission in a way that was generally consistent with published guidelines. In particular, most considered of paramount importance the prognosis of the acute illness and of the underlying disease. These two factors belong to the basic principles guiding triage, and earlier studies have shown that they are associated with admission to intensive care.5^–7

Patients' wishes were considered important by most respondents, and in the experimental part of the study were strongly related to the decision to admit. This reflects a departure from the traditional paternalistic attitude and the increasing attention paid to self determination and autonomy.17

This sharply contrasts with the finding that the patient's personality equally influenced doctors' decisions. The preference to admit patients with a positive attitude was not conscious as respondents rated the patient's emotional state low among the determinants for admission, which agrees with recommendations that a patient's personal and behavioural characteristics should not influence triage.3 Similarly, a patient described as socially active was more likely to be admitted. These results raise ethical concerns.2 3 Doctors seemed to assign a certain worth to the patient based on subjective and mainly unconscious criteria. Unfair allocation of resources may result. This attitude can reflect both individual and broader western cultural values.18

Availability of beds was considered important by the respondents and was correlated with the decision to admit in the vignettes. Previous studies have suggested that fewer patients are admitted when beds are scarce.6 Our study shows that this is the case. Doctors also recognised different aspects of rationing such as nursing workload and the optimal use of available beds.19

Age was moderately associated with admission to intensive care. Age is in itself probably not an important determinant and is balanced against medical and contextual factors.9

Doctors' decisions varied according to the family's wishes and attitudes. An explicit request was related to the decision to admit to intensive care but apparent emotional attachment was not. This raises the issues of whether the patient's best interests were the doctor's prime concern in both situations and whether the doctor should actively seek the preferences of the family. Decision making in the presence of an incompetent patient is problematic. The accuracy of proxy judgments is generally poor,20 but doctors' predictions of the patient's preferences are even worse.21 Family members can also exert undue pressure on the doctor,22 and they often fail to fully understand their relative's state of health and related issues.23

The main limitation of our study is the possible discrepancy between doctors' decisions in practice and their answers to vignettes with hypothetical patients. However, the validity of the decisions is supported by the overall admission rate of about 50% and the respondents' straightforward answers, as exemplified by the consideration given to the availability of beds. The high admission rate of the patient with relapsing leukaemia may seem surprising, but the respondents' comments about having little choice because the patient was mechanically ventilated, further indicate that their answers were based on their everyday practice. A previous study found no correlation between relapsing leukaemia with severe acute respiratory failure and admission to intensive care.5 The response rate of 61% raises the possibility of non-response bias, particularly for descriptive variables such as the rating of factors influencing decisions on admission. Because response rates were similar for the eight versions of the questionnaire, a bias is unlikely in measures of association between the experimentally manipulated factors and the reported decisions to admit patients.

In conclusion, doctors do not discriminate against patients with cancer when deciding on admission to intensive care. The decision making process is influenced by the patient's wishes and ethically questionable factors such as the patient's personality. The medical community must be aware of the existence of unconscious value judgments leading to possible biased decisions for admissions to intensive care.