Bedside perfusion tests diverge in ICU patients, capillary refill time best predicts mortality

Peripheral perfusion is increasingly recognized as a key determinant of outcomes in critically ill patients, yet the relationships among the common bedside tools used to assess it have remained unclear. A prospective pilot study conducted in a general 34-bed intensive care unit at the Hôpital Universitaire de Bruxelles in Belgium set out to map how six widely used parameters behave when measured simultaneously in the same patients. The findings, published in Annals of Intensive Care, suggest that most of these tools capture overlapping physiological signals, while one in particular, capillary refill time, stood out as the most reliable marker of ICU mortality.

Traditional resuscitation strategies focus on macrocirculatory targets such as blood pressure and cardiac output. However, restoring these parameters does not guarantee adequate tissue oxygenation. Alterations in the microcirculation are common in critical illness and have been linked to organ failure and poor prognosis. Importantly, microcirculatory dysfunction can occur even in the absence of overt macrocirculatory abnormalities, underscoring the need for a more integrated hemodynamic approach. Improvements in microcirculatory perfusion have been linked to better organ function and survival, highlighting its relevance as a therapeutic target.

A variety of techniques are currently available to assess microcirculatory perfusion, but no single tool provides an exhaustive and simultaneous assessment of all key microcirculatory parameters, including capillary density, heterogeneity and microvascular flow, vasoreactivity, capillary recruitment, and the efficiency of oxygen diffusion and uptake at the tissue level. Some studies have focused on clinical evaluation such as the mottling score and capillary refill time, while others have used instrumental assessment including laser doppler flowmetry, plethysmography, near-infrared spectroscopy, or direct visualization of microvascular flow using handheld video-microscopy. Whether these various tools provide convergent or divergent assessments of peripheral perfusion is poorly studied, as comparative studies are scarce and in most cases only evaluate two techniques at a time.

The study included adult ICU patients whose expected length of stay was at least three days, measured after initial resuscitation without significant hemodynamic variations. On day 1, six peripheral perfusion parameters were simultaneously measured: the Peripheral Perfusion Index (PPI), the Mottling Score (MS), Capillary Refill Time (CRT), the central-to-peripheral temperature gradient (ΔT), Skin Blood Flow at basal temperature (SBFBT), and forearm tissue oxygenation (rSO₂). The primary outcome was the correlation between these parameters, while secondary outcomes included subgroup consistency and associations with ICU mortality. Sensitivity analyses evaluated the concordance of changes in these parameters over the first three days of ICU stay.

Fifty-five patients were included, with a median age of 64 and 65.5 percent male. Circulatory shock was the leading admission cause at 36.4 percent, followed by acute brain injury at 29.1 percent and acute respiratory failure at 25.4 percent. The median length of stay in the ICU was 5 days. All day 1 peripheral perfusion parameters were collected at a median time of 18.5 hours after ICU admission. On day 1, the median heart rate was 87 bpm, mean arterial pressure was 78 mmHg, and lactate level was 1.7 mmol/L.

The correlation analysis revealed strong relationships among several parameters. PPI showed a strong positive correlation with SBFBT (ρ = 0.72) and a strong negative correlation with ΔT (ρ = -0.68). CRT was strongly negatively correlated with SBFBT (ρ = -0.66) and positively correlated with ΔT (ρ = 0.54). CRT and PPI showed a moderate negative correlation (ρ = -0.52), while SBFBT exhibited a strong negative correlation with ΔT (ρ = -0.80). Notably, rSO₂ did not show significant correlations with any of the other variables, suggesting that it may reflect different, potentially complementary, physiological aspects of tissue perfusion.

In subgroup analyses, the same associations were observed and strengthened in patients with circulatory shock. In patients with acute brain injury, only the association between SBFBT and ΔT was maintained, while in patients with acute respiratory failure, associations were maintained between ΔT and PPI, and between ΔT and SBFBT. The observed loss of coherence between peripheral perfusion parameters in patients with acute brain injury may reflect an early alteration of systemic microcirculatory autoregulation, as previously suggested in preclinical models. Two experimental studies reported both impaired endothelium-dependent vasodilation and dysregulated microcirculatory blood flow following cerebral injury, supporting the hypothesis that systemic microcirculatory disturbances occur in this population and may be dissociated from global hemodynamics.

Peripheral perfusion parameters on admission differed between survivors and non-survivors. Among the 14 non-survivors, PPI and SBFBT were significantly lower than in survivors, the mottling score was higher, and CRT was longer. No significant difference was observed for ΔT and rSO₂ values between the two groups. Using receiver operating characteristic curve analysis, CRT demonstrated the highest predictive value for ICU mortality in the overall population, with an area under the curve of 0.75 (95% CI: 0.62–0.88, p = 0.007). The mottling score also showed good discriminative ability with an AUC of 0.72, while SBFBT and PPI were significant predictors with AUCs of 0.69 each. ΔT and rSO₂ were not significant predictors. In the circulatory shock group, none of the parameters demonstrated significant discriminatory ability, likely due to greater heterogeneity in CRT values despite apparent hemodynamic optimization. In patients with acute brain injury, CRT showed the highest predictive value, with an AUC of 0.92.

Longitudinal analyses showed that some correlations remained significant over time, especially between SBFBT and ΔT, which maintained a strong negative correlation (ρ = -0.67). PPI and CRT also maintained significant correlations with these two parameters, while rSO₂ showed no significant correlation with the other parameters over time. This suggests that either SBFBT or ΔT could be used for microcirculatory monitoring over time, depending on their availability in the clinical setting.

The authors emphasize that the study was designed as an exploratory, hypothesis-generating work to assess the feasibility and potential relevance of multimodal peripheral perfusion assessment in a mixed ICU population. Its small sample size and monocentric design inherently limit the statistical power and generalizability of the findings, particularly in subgroup analyses. The daily monitoring approach may have missed short-term fluctuations in tissue perfusion, and the lack of dynamic assessment such as a vascular occlusion test may have limited the ability of rSO₂ measurements to reflect microcirculatory function. The potential influence of skin pigmentation on clinical assessment tools such as CRT and the mottling score cannot be excluded, and the study was unable to compare these measurements to the current gold standard of sublingual microcirculation imaging. Based on the data, the authors cannot suggest that peripheral perfusion indices are interchangeable, and no single tool appears able to provide a comprehensive assessment of microcirculatory function in critical illness. Bedside interpretation of all these tools should be adapted to the patient’s clinical phenotype, disease severity, and predominant circulatory derangement to optimize their diagnostic and prognostic value.

Subject of Research: Correlation and prognostic value of bedside peripheral perfusion parameters in critically ill patients

Article Title: Multimodal assessment of peripheral perfusion in critically ill patients: a pilot study

Article References: Demailly, Z., Modica, E., Vitali, E., Sousa, F., Dragoni, C., Zernini, I. S., Annicchiarico, C., Dehout, C., Tamion, F., Besnier, E., Njimi, H., dos Santos, E. C., & Taccone, F. S. (2025). Multimodal assessment of peripheral perfusion in critically ill patients: a pilot study. Annals of Intensive Care, 15(1), Article 176. https://doi.org/10.1186/s13613-025-01585-2

Image Credits: AI Generated

DOI: 10.1186/s13613-025-01585-2

Keywords: peripheral perfusion, capillary refill time, mottling score, near-infrared spectroscopy, circulatory shock, acute brain injury, acute respiratory failure, ICU mortality, microcirculation, skin blood flow, temperature gradient, pilot study