To our knowledge, this is the largest point prevalence study of mobilisation practices in critical care outside of the COVID-19 pandemic. Forty percent of patients mobilised out of bed, higher than the 24–36% range previously published [15,16,17,18,19]. Out-of-bed mobility was associated with a RASS within the range − 1 to + 1, lower doses of vasopressors and inotropes, lower FiO2.

While some variables were statistically significant between mobilised and not-mobilised groups, there was considerable overlap of the distribution of these variables. Of note is the lack of clinically meaningful separation of any of the key parameter values.

The majority of patients who met the green risk criteria were mobilised. However, of the those in critical care for less than 72 h, 60% who met the green criteria were elective surgery patients. Only 39% of non-elective patients met green criteria, while 50% fulfilled red criteria. Higher mobility levels in elective surgical patients were also reported in an Australian cohort study [24]. As such, studies focusing solely on elective surgical patients should not be generalised to the broader critical care population. There may be scope for increasing mobility levels in the green non-mobilised group. The Hodgson et al. [14] consensus focused on safe parameters for mechanically ventilated patients. However, the non-mobilised cohort also included those not receiving mechanical ventilatory support. The presence of an artificial airway (whether endotracheal tube or tracheostomy) is classified as a ‘green’ risk characteristic; however, the recommendations specify taking the most conservatively scored individual parameter as group-membership (i.e. any single amber or red), and specifically comment that their aim is to provide guidance to maximise safe mobilisation [14] For these reasons, and the absence of other high-quality guidance on safe mobilisation of critical care patients, the risk rating was applied to the whole cohort in this study.

The wide confidence interval around both the VIS and each individual vasoactive drug suggests that clinical decision making is not based solely on an arbitrary dose, but is dependent on individual patient physiology. The observation that the threshold for mobilising was a VIS of 10 should not be interpreted that it is safe to mobilise all patients with scores < 10. There may be a cohort of patients receiving vasoactive support in whom the cumulative risk of other factors precluding mobilisation is deemed acceptable by those taking the decision to mobilise.

Commentators of critical care mobility practices often report low levels of mobility as problematic. Although these data showed that only 40% of all patients were mobilised, 89% and 75% of those that met green and amber risk ratings (i.e. the only published international consensus on safety of out-of-bed activity) mobilised, respectively. Re-interpreting the data in the context of physiological safety paints a different picture, suggesting 90% of stable patients achieved an appropriate mobility goal within 72 h.

Perhaps there is a time during early severe acute illness when mobilising is neither safe nor feasible. Of note, the ‘early mobility' study by Schweickert et al. [2] reported a significantly shorter time to death in the intervention group. Similarly, the randomised control trial by Schaller and colleagues [10] of early goal-directed mobilisation in a surgical critical care unit showed improved physical function in the intervention group, but a doubling in mortality rate compared to the control limb. The reduction in critical care length of stay is likely to have been confounded by the earlier deaths in the intervention group. A recent multi-centre randomised controlled trial of 750 patients showed no benefit of protocolised early mobilisation versus usual care in terms of morbidity, disability or activities of daily living [13] but an increase in adverse events. This raises the important question of the potential iatrogenic impact of early intensive mobilisation in those with critical illness, and a ceiling effect of dose-benefit, as demonstrated in other populations such as acute stroke [25].

There was a marked physiological variability within groups. Those not mobilised had a pattern of marginally poorer physiological status, but no markers alone met criteria of clinical significance. Hence, each patient’s physiological status is unique, and decision-making is necessarily individualised.

We acknowledge that this observational study has several limitations. Adverse events were not specifically collected, but none were recorded narratively as reasons why a greater level of mobility was not achieved. The 960 patients represent only 16% of the potential critical care capacity in the UK and cover only 20% of the acute health and social care NHS organisation in England, Wales and Northern Ireland, but do represent both specialist and non-specialist hospitals. The data collection on the majority of the sites was carried out by physiotherapists who were part of the patients care team, which may have resulted in less capacity to mobilise patients, potentially reflected in the 30 (3%) patients not mobilising due to staffing issues. However, we believe that this cohort study reflects current practice across the UK.

Variation in usual care altering the dose–effect of mobilisation or other physical rehabilitation interventions is often discussed as a limitation of research in this field. From the data collected in the present study, it could be argued that the lack of well-defined usual care in mobilisation, and significant variation within controls patients, limit the generalisability of the research published to date. Future trials need a formal control group with well-defined usual care in order to make meaningful comparisons with interventions of altered type, dose, timing, or intensity of physical rehabilitation.