We have found a population prevalence of adult critical illness per 100,000 people of 19.4 in the Sörmland Region of Sweden. Of all hospitalized patients, 10.5% were critically ill and of these critically ill, 94% were cared for in general wards. To our knowledge, this is the first attempt to estimate the burden of critical illness at population level using prospective enrolment and examination of patients.
The prevalence of critical illness
This burden of critical illness is higher than previously considered, but comparisons with previous estimates are challenging. Since the burden of acute illnesses is often measured using the incidence, (the flow of cases into the system per time unit), a conversion of our prevalence estimate is needed. Using an assumption that critical illness on average has a duration of 1.1 days, (based on the proxy of the median duration of patients’ care in Swedish ICUs)[25], our findings would correspond to 17.6 new cases of critical illness per 100,000 people per day. This number is substantially higher than the estimate by Adhikari and colleagues of 1.3-1.9 per 100,000 people per day in Europe.[2] The difference could be explained by our more liberal criteria for critical illness rather than limiting inclusion to patients with certain syndromes and those treated in ICUs. We don’t believe our criteria were too liberal. Severely abnormal vital signs have been shown to identify a group of patients with a high risk of death – indeed their risk of 26-28% 30-day mortality[13] is higher even than the risk for patients admitted to ICUs (17%)[19]. The burden of critical illness in our study is also larger than the burden of major acute diagnoses that receive substantial public, research, and policy interest. There are 0.5 strokes[26], 2.7 hip fractures [27], and 2.8[28] myocardial infarctions in Sweden per 100,000 people per day, with 30-day mortalities of 11%[26], 8%[29], and 8%[30] respectively. Critical illness appears to be a larger health issue than is usually recognized.
There are important policy implications of this large burden of critical illness. The health system requirements need to be specified for these patients, as do the competencies of health workers to care for them, and the clinical systems and routines to prioritize[31, 32] for them to be quickly identified as critically ill and promptly treated – even when resources are limited such as at 3am on a Sunday night[33]. Re-designing health services to have an increased focus on critical illness, in the same way as the longstanding focus on diagnoses with high mortalities[34-36], would target the highest risk patients and – as they are common – could be favorable for quality improvement, research, and innovation. Improving care of severely ill patients would benefit patients suffering from any underlying condition and have the advantage of improving care for patients lacking a definitive diagnosis, those with multimorbidity and those who have been mis-diagnosed.
The proportion of patients in hospital that are critically ill
Among all in-patients, 10.5% were critically ill, a finding in-line with previous research. Studies from university hospitals in Finland and Sweden report proportions of patients with single parameter signs of 8.4 and 12% and NEWS≥7 of 6.0 and 6.5% respectively[13, 37], In our population-based study, most patients were treated in district hospitals, and of these patients 10% were critically ill (versus 13% in the university hospitals). While it may be thought that most patients with critical illness would be referred to ICUs and to larger hospitals, the findings do not support that perception.
The in-hospital location of critically ill patients
A large majority of critically ill patients, (94%), were cared for in general wards. There are, to our knowledge, no previous studies of the in-hospital location of critically ill patients, but the presence of very ill patients in general wards and not just in HDUs and ICUs has been described in work on sepsis, early warning systems (EWS), rapid response teams (RRT) – and from low-resource settings[16, 38-43]. The implications of this finding are debatable[44]. An argument could be proposed that an ICU-bed capacity corresponding to that in the USA and Germany (>20 per 100,000 people) could be needed to care for these patients. However, from a health system perspective, it would be more rational to provide care at the lowest effective level, to maximize the use of resources[45, 46]. ICUs may offer better care in some cases[47], but at a high cost [48] – and many countries in the world have less than one ICU bed per 100,000 people[49]. ICU care did not provide better outcomes for patients lacking a strong ICU indication and was not always cost-effective in the COVID-pandemic[46, 50]. For a large proportion of the critically ill, HDUs providing less advanced critical care together with optimized care of critical illness in general wards may be a better approach to prevent deterioration, the need for ICU care and death[46, 51]. This requires that health systems ensure adequate provision of fundamental critical care in the general wards, such as through training of staff, the use of EWS and RRT, and focused quality improvement[51-53]. A recent consensus has specified ‘Essential Emergency and Critical Care’, forty clinical processes that are effective, lifesaving and feasible to deliver in general wards and all other parts of hospitals[33].
Strengths and limitations
The strengths of the study include the use of criteria to define critical illness that are neither dependent on the underlying diagnosis nor the hospital location; the prospective examination of all study participants by trained data collectors and quality secured equipment; the few missing data points; and the high inclusion rate of patients from all types of wards and hospitals in which a well-defined population are admitted for care.
The study had limitations. First, the pragmatic vital signs-based criteria used may have missed some high-risk patients whose vital signs were insufficiently deranged or had been stabilized through care. Conversely, some patients with adapted physiology due to chronic disease may have been classified as being critically ill. Second, for logistical reasons, we could neither include patients in the operating theatres nor patients in the emergency units awaiting a decision about hospital admission, some of whom may have been critically ill. Third, data collections took place before the COVID-19 pandemic, in the daytime on weekdays and the prevalence may differ over time and between weekends and nights[54]. Fourth, the prevalence estimates changed when alternative criteria for critical illness were used in the sensitivity analyses, highlighting the challenge of identifying critical illness and the need for a process towards agreed criteria for critical illness[3]. Fifth, we assumed that there is no critical illness in the community. If this assumption is false, the burden of critical illness may be greater than our estimate. Lastly, this study was from one region in one country. However, we have no reason to believe that there are substantial differences between settings similar to Sweden, and a transfer of the findings can be useful for many health systems.