Pleth variability index as a non-invasive prognostic indicator in acute exacerbation of chronic obstructive pulmonary disease

Abstract

Background

Early risk stratification in Chronic Obstructive Pulmonary Disease (COPD) exacerbations is essential for effective management. The Pleth Variability Index (PVI), a non-invasive parameter derived from photoplethysmographic waveforms, has been evaluated in various clinical settings; however, its relevance to COPD exacerbation severity in adults remains unclear. This study aimed to assess the association between admission PVI values and COPD exacerbation severity, hospital admission, and 30-day mortality.

Methods

This prospective, single-center study was conducted in a tertiary hospital between September 2023 and September 2024 with ethics approval. Adults (≥ 18 years) with COPD exacerbation were included, while those with cardiac or other pulmonary conditions, or factors affecting PVI, were excluded. PVI was measured non-invasively at admission and 2 h post-treatment using the Masimo Root device. Clinical data, laboratory results, outcomes, and 30-day mortality were recorded. COPD exacerbation severity was classified according to the Global Initiative for Obstructive Lung Disease (GOLD), and patients were grouped by their outcomes. Analyses included Receiver operating characteristic (ROC) curves, correlation, and regression models.

Results

Of 204 patients, 131 were included (mean age 70.6 ± 10.5; 77.1% male). Exacerbations were mild in 23 (17.6%), moderate in 49 (37.4%), and severe in 59 (45.0%). The median PVI was 42% at admission and 27% post-treatment (p < 0.001), with higher values observed in patients with more severe cases. 34 (25.9%) discharged, 38 (29.0%) admitted to ward, 59 (45.0%) to Intensive Care Unit (ICU), and 17 (13.0%) died. ICU and mortality groups had higher admission PVI. A baseline PVI > 40 predicted severe exacerbation (AUC = 0.828), ICU admission (AUC = 0.867), and mortality (AUC = 0.774). PVI correlated negatively with oxygen saturation and pH, positively with Partial Pressure of Carbon Dioxide (pCO₂), and was an independent predictor of ICU admission and 30-day mortality.

Conclusion

PVI emerges as a valuable parameter in predicting exacerbation severity, hospital admission, and mortality in COPD exacerbations. With non-invasive and rapidly measurable advantages, PVI can be utilized as a complementary prognostic tool in triage and treatment processes in the emergency department.

Background

Chronic Obstructive Pulmonary Disease (COPD) is a common respiratory condition characterized by irreversible airflow limitation and represents a significant cause of morbidity and mortality worldwide [1]. COPD exacerbations are acute clinical events, often triggered by infections or environmental factors, that contribute to disease progression, impaired quality of life, and increased healthcare utilization [2]. Early risk stratification during exacerbations is crucial for guiding treatment strategies and optimizing the use of hospital resources [1].

During obstructive airway diseases, increased respiratory effort and negative intrathoracic pressure during inspiration may lead to enhanced right ventricular filling and reduced left ventricular filling, generating a hemodynamic response similar to pulsus paradoxus (PP) [3]. Traditionally, PP is defined as a decrease of ≥ 10 mmHg in arterial systolic pressure during inspiration; however, these hemodynamic changes can also be observed in photoplethysmographic waveforms [4, 5]. PVI is a non-invasive parameter calculated by assessing fluctuations in the perfusion index over at least one respiratory cycle [6]. Previous studies have demonstrated that PVI is a reliable non-invasive method for evaluating PP [7, 8]. In this context, PVI measurement may help determine disease severity in patients presenting with asthma or COPD exacerbations.

This study aimed to evaluate the association between PVI values measured at the time of admission in patients presenting to the emergency department (ED) with COPD exacerbation and the severity of exacerbation, hospital admission, and mortality, and to investigate the potential of PVI levels in predicting adverse clinical outcomes.

Methods

Study design and ethics approval

This study was designed as a single-center, prospective clinical investigation conducted in a tertiary care hospital. The study was conducted in accordance with the Declaration of Helsinki and was approved by the local ethics committee (Protocol No: SÜKAEK/2023/17/15, Date: 18 September 2023). The study included patients diagnosed with COPD who presented to the ED between September 2023 and September 2024.

Study population

Patients aged 18 years or older with a known diagnosis of COPD, presenting with respiratory symptoms such as cough, dyspnea, sputum production, and wheezing, and diagnosed with acute COPD exacerbation were eligible for inclusion. Written informed consent was obtained from all patients. In cases where patients were unable to sign, written consent was obtained from their legal guardians. Exclusion criteria included: age under 18 years, diagnosis of acute coronary syndrome, heart failure, recurrent ED visits within one week, and other pulmonary conditions such as pneumothorax, pulmonary edema, or pulmonary embolism. Additionally, patients with clinical conditions likely to affect PVI values (e.g., septic shock, hemorrhagic shock, or dehydration) were excluded; dehydration and sepsis/septic shock were determined using standard clinical assessment at presentation, and sepsis and septic shock were defined according to Sepsis-3 criteria [9].

PVI measurement and patient management

PVI was measured noninvasively and automatically calculated by a commercially available pulse oximeter (Masimo Radical-7^®, Masimo Corporation, Irvine, CA, USA). PVI is a dynamic parameter that noninvasively measures peripheral perfusion variability and is used to assess fluid responsiveness. PVI is based on the analysis of amplitude variations in the photoplethysmographic (pleth) waveform during the respiratory cycle. The measurement is performed by evaluating changes in light absorbance as infrared light passes through the tissue [10–12].

PVI measurements were obtained using the Root-Masimo monitor (Masimo Corporation, Irvine, CA, USA). The device processes photoplethysmographic signals acquired by sensors to determine Perfusion Index (PI) values and automatically calculates the PVI based on the respiratory-induced maximum and minimum variations in PI [10].

The following formula defines PVI:

Here, PImax and PImin represent the maximum and minimum perfusion index values observed during the respiratory cycle, respectively. These variations reflect the influence of intrathoracic pressure changes on peripheral circulation.

In this study, all PVI measurements were performed by an investigator who was not involved in the patients’ clinical management. The PVI values were recorded and stored electronically without being displayed on the bedside monitor or disclosed to the treating physicians, nursing staff, or other members of the clinical team. Consequently, all treatment decisions were made independently of PVI data to minimize performance bias and preserve the objectivity of outcome assessment.

PVI was measured twice: (1) at ED admission, immediately upon presentation and before initiation of ED treatment, and (2) two hours after standardized therapy. The type of respiratory support in use at each measurement time point (room air, conventional oxygen supplementation, or non-invasive ventilation) was recorded. During this interval, all patients received conventional management in accordance with the Global Initiative for Obstructive Lung Disease (GOLD) guidelines. Patients were managed in the ED [1], including controlled oxygen therapy, short-acting bronchodilators, systemic corticosteroids, and non-invasive ventilation when clinically indicated.

Data collection and patient grouping

Demographic characteristics, comorbidities, vital signs, laboratory results (including arterial blood gases [ABG] and C-reactive protein [CRP]), exacerbation severity, and 30-day mortality data were recorded. Exacerbation severity was operationally categorized using GOLD-based treatment-intensity criteria:

Mild exacerbation: managed with increased short-acting bronchodilators only, without systemic corticosteroids and without antibiotics, and not requiring an ED visit or hospitalization. Moderate exacerbation: requiring systemic corticosteroids and/or antibiotics but not requiring hospitalization. Severe exacerbation: requiring ED presentation and/or hospital admission, including need for ventilatory support (non-invasive or invasive), Intensive Care Unit (ICU) admission, or management of acute respiratory failure [1]. Patients were also categorized according to clinical outcomes (discharged, admitted to the ward, admitted to the ICU) and 30-day mortality. Comparative statistical analyses were performed across these groups.

Statistical analysis

Data were analyzed using IBM SPSS Statistics software (version 25; IBM Corp., Armonk, NY, USA) and MedCalc software (version 20; MedCalc Software Ltd., Ostend, Belgium). A 95% confidence interval was applied in all analyses. Normality of distribution was assessed using the Kolmogorov-Smirnov test. Normally distributed numerical variables were expressed as mean ± standard deviation, while non-normally distributed variables were expressed as median (minimum–maximum). Categorical variables were presented as frequency (n) and percentage (%).

Comparisons between two independent groups were performed using the Student’s t-test for normally distributed variables and the Mann-Whitney U test for non-normally distributed variables. The chi-square test or Fisher’s exact test was applied for categorical data. Pairwise comparisons were adjusted using the Bonferroni correction. Pre- and post-treatment analyses were conducted using the Wilcoxon signed-rank test for continuous variables and the marginal homogeneity test for categorical variables.

Receiver Operating Characteristic (ROC) curve analysis was used to determine the optimal cut-off values of PVI for diagnostic and prognostic purposes, including sensitivity and specificity. The area under the curve (AUC) and 95% confidence intervals (CI) were reported. Correlations between PVI and other parameters such as oxygen saturation, pH, and pCO₂ were evaluated using Spearman correlation analysis. Comparative logistic regression models (ABG-only vs. ABG + PVI) were constructed for ICU admission and 30-day mortality, and model discrimination was compared using ROC–AUC analysis with DeLong testing. A p-value < 0.05 was considered statistically significant.

Results

During the study period, 204 patients presented to the ED with respiratory symptoms. Seventy-three patients who met at least one exclusion criterion were excluded. A total of 131 patients diagnosed with COPD exacerbation were included in the study. The clinical and demographic characteristics of the patients are presented in Table 1.

Table 1.

Baseline characteristics, vital signs and clinical parameters of the study population

Properties/Variables		Values
Number of Patients		131
Age		70.57 ± 10.48
Gender	Female	30 (22.9%)
	Male	101 (77.1%)
Vital Signs	Systolic Blood Pressure (mmHg)	142.35 ± 28.46
	Diastolic Blood Pressure (mmHg)	77.25 ± 15.45
	Mean Arterial Pressure (mmHg)	98.95 ± 17.48
	Pulse (beats/min)	102.9 ± 21.66
	Body Temperature (°C)	36.6 (36.0-39.8)
	Respiratory Rate (/min)	25 (18–60)
SaO2 (%)	Before treatment	83 (44–96)
	After treatment	94 (78–100)
pH	Before treatment	7.33 (7.08–7.56)
	After treatment	7.37 (7.21–7.45)
pCO2	Before treatment	55.2 (33.2–120.0)
	After treatment	47.9 (29.8–81.7)
CRP		52.3 (0.9-483.8)
PVI (%)	Before treatment	42 (14–99)
	After treatment	27 (8–68)
Exacerbation severity	Mild	23 (17.56%)
	Moderate	49 (37.40%)
	Severe	59 (45.04%)
Outcome	Discharged	34 (25.95%)
	Admitted to the ward	38 (29.01%)
	Admitted to the ICU	59 (45.04%)
Mortality		17 (12.98%)

SaO₂ Oxygen saturation, pCO₂ Partial pressure of carbon dioxide in arterial blood gases, CRP C-reactive protein, PVI Pleth Variability Index, ICU Intensive care unit

Of the COPD patients, 23 (17.56%) presented with mild exacerbations, 49 (37.40%) with moderate exacerbations, and 59 (45.04%) with severe exacerbations. The median pre-treatment PVI value was 42% (14–99), while the median post-treatment PVI value was 27% (8–68). Following completion of initial treatment in the ED, 34 patients (25.95%) were discharged, 38 (29.01%) were admitted to the general ward, and 59 (45.04%) were admitted to the ICU. Within 30 days of admission, mortality occurred in 17 patients (12.98%).

When patients were classified according to pre-treatment disease severity, there were no significant differences in age and gender among the mild, moderate, and severe groups (p = 0.917 and p = 0.783, respectively). Both pre-treatment and post-treatment PVI values were higher in the severe exacerbation group compared to the mild and moderate groups, and higher in the moderate group compared to the mild group (p < 0.001) (Fig. 1). Post-treatment PVI values were significantly lower than pre-treatment values across all groups (p < 0.001). Similarly, pre-treatment oxygen saturation values were significantly lower than post-treatment values in all groups (p < 0.001) (Table 2).

Fig. 1.

Comparison of median PVI values before and after treatment by exacerbation severity, outcome, and 30-days mortality (PVI: Pleth variability index, ICU: Intensive care unit)

Table 2.

Comparison of pretreatment and posttreatment PVI values by exacerbation severity and patient outcomes

		Pretreatment PVI	Posttreatment PVI	p
Exacerbation Severity	Mild (n=23)	24 (14-45)a	15 (8-55)a	0.038
	Moderate (n=49)	36 (14-69)b	25 (8-49)b	<0.001
	Severe (n=59)	50 (31-99)c	31 (11-68)c	<0.001
p (between severity groups)		<0.001	<0.001
Outcome	Discharged (n=34)	32 (14-71)a	18.5 (8-55)a	<0.001
	Admitted to ward (n=38)	31.5 (17-52)a	23 (8-46)a	<0.001
	Admitted to ICU (n=59)	52 (27-99)b	33 (11-68)b	<0.001
p (between outcome groups)		<0.001	<0.001

PVI Pleth Variability Index, ICU Intensive care unit

^a,b,cShows statistically significant differences between severity and outcome groups

When patients were grouped according to clinical outcomes, there were no significant differences in age and gender among those discharged, admitted to the ward, or admitted to the ICU (p = 0.054 and p = 0.720, respectively). No significant differences were found in pre- and post-treatment PVI values between discharged patients and those admitted to the ward. In contrast, patients admitted to the ICU had significantly higher PVI values than both discharged patients and those admitted to the ward (p < 0.001) (Fig. 1). Post-treatment PVI values were significantly lower compared to pre-treatment values within each group (p < 0.001) (Table 2).

In patients who died, median pre-treatment and post-treatment PVI values were 55 (30–99) and 32 (16–46), respectively. In survivors, these values were 38.5 (14–77) and 27 (8–68), respectively. Pre-treatment PVI values were significantly higher in patients who died compared to survivors (p < 0.001). In contrast, post-treatment PVI values were not statistically different between survivors and non-survivors (p = 0.074) (Fig. 1).

The diagnostic performance of PVI in predicting severe and moderate exacerbations, ICU admission, and mortality was evaluated using ROC curve analysis. The areas under the ROC curves (AUCs) for PVI were 0.828, 0.894, 0.867, and 0.774, respectively (Table 3; Fig. 2).

Table 3.

Diagnostic performance of PVI in predicting clinical outcome

Cut-off	Severe Exacerbation	ICU Admission	Mortality
	> 40	> 41	> 40
Sensitivity	86.44%	86.44%	94.12%
Specificity	75.00%	79.17%	53.51%
Accuracy	80.15%	82.45%	58.78%
PPV	73.9%	77.3%	23.2%
NPV	87.1%	87.7%	98.4%
AUC (95% CI)	0.828 (0.753–0.889)	0.867 (0.797–0.920)	0.774 (0.692–0.842)
p	< 0.001	< 0.001	< 0.001

ICU Intensive Care Unit, PPV Positive predictive value, NPV Negative predictive value, AUC Area under the curve, CI Confidence interval

Fig. 2.

ROC curves of PVI for predicting clinical outcomes (ICU: Intensive care unit, AUC: area under the curve)

A negative correlation was observed between pre-treatment PVI values and pre- and post-treatment oxygen saturation and pH values. In contrast, a positive correlation was found with pre- and post-treatment pCO₂ values (p < 0.001). Univariate and multivariate regression analyses identified elevated PVI as an independent risk factor for both ICU admission (p < 0.001) and 30-day mortality (p = 0.002) (Table 4).

Table 4.

Univariate and multivariate logistic regression results for ICU admission and 30-day mortality

	ICU Admission				30-days Mortality
	Univariate		Multivariate		Univariate		Multivariate
	OR	p	Adj OR	p	OR	p	Adj OR	p
Age	1.030	0.089			1.044	0.085
Gender (male)	0.918	0.838			0.960	0.947
pH	8.818	< 0.001	0.0003	0.190	0.036	0.366
SaO2	0.894	< 0.001	0.950	0.137	0.955	0.043	0.997	0.908
PCO2	1.118	< 0.001	1.071	0.022	1.008	0.644
PVI	1.132	< 0.001	1.121	< 0.001	1.075	< 0.001	1.072	0.002
CRP	1.006	0.012	1.006	0.071	1.005	0.033	1.004	0.159

ICU Intensive care unit, OR Odds ratio, SaO₂ Oxygen saturation, pCO₂ Partial pressure of carbon dioxide in arterial blood, PVI Pleth Variability Index, CRP C-reactive protein

The baseline logistic regression model including age, gender, SaO₂, pH, pCO₂, and CRP yielded an AUC of 0.875 (95% CI: 0.805–0.926) for predicting ICU admission. After adding PVI, the AUC increased to 0.938 (95% CI: 0.882–0.973). The DeLong method confirmed this improvement was statistically significant (ΔAUC = 0.0633; 95% CI: 0.0151–0.112; z = 2.576; p = 0.010), emphasizing the added value of PVI in predictive modeling.

For 30-day mortality, the baseline model including age, gender, SaO₂, pH, pCO₂, and CRP yielded an AUC of 0.696 (95% CI: 0.609–0.773). After adding PVI to the model, the AUC increased to 0.807 (95% CI: 0.728–0.870). Although the improvement in discriminative performance did not reach statistical significance (ΔAUC = 0.111; 95% CI: −0.0186 to 0.240; z = 1.679; p = 0.093 by DeLong test), the addition of PVI resulted in a numerically relevant increase in AUC.

Discussion

This study represents the first prospective investigation to evaluate the relationship between PVI values measured at admission and after treatment in patients presenting to the ED with COPD exacerbation, and their association with exacerbation severity, clinical course, treatment response, and outcome parameters. Our findings demonstrated that PVI values measured at admission increased significantly with exacerbation severity, and statistically significant differences were observed among the groups. Moreover, both pre- and post-treatment PVI values were significantly associated with clinical outcomes such as hospitalization and mortality. To the best of our knowledge, current literature includes only pediatric studies examining the association between PVI and asthma exacerbation severity in obstructive airway diseases [6–8], and no study has investigated this association in the adult population with COPD exacerbations. In this context, our study provides novel evidence highlighting the potential role of PVI as a prognostic and predictive marker in COPD exacerbations, adding originality and value to the existing literature.

Although asthma and COPD are generally classified as distinct conditions, recent evidence demonstrates that they share important pathophysiological mechanisms. Both diseases are characterized by chronic airway inflammation, obstruction, and airway remodeling [13]. PVI, initially described as a non-invasive hemodynamic parameter for predicting fluid responsiveness in mechanically ventilated patients, has been widely studied in various clinical contexts, including intraoperative fluid management, evaluation of the effects of mechanical ventilation on cardiac output, and prediction of mortality in critically ill patients [14–18]. Furthermore, PVI has been reported as a promising parameter for rapidly assessing disease severity in children with asthma or obstructive airway diseases [7]. A prospective study by Demir et al. found a strong relationship between admission PVI values and clinical severity in children with obstructive airway disease, with significant reductions in PVI after treatment [6]. In acute COPD exacerbations, dynamic hyperinflation and intrinsic PEEP can increase intrathoracic pressure excursions and ventricular interdependence, leading to cyclic changes in venous return and left ventricular filling; these heart-lung interactions may amplify respiratory variations in peripheral plethysmographic signals and thereby influence PVI values [19–21].

Similarly, our study observed significant associations between PVI values, exacerbation severity, and clinical outcomes in adult patients with COPD exacerbation. These findings suggest that PVI may provide predictive value in pediatric populations and adult patients with COPD.

Our results showed that PVI values at both admission and post-treatment were higher in patients admitted to the ICU compared with those admitted to the ward or those who were discharged. In addition, patients who died had significantly higher baseline PVI values compared with survivors. Particularly, PVI values greater than 40% at admission were able to identify patients with severe exacerbations, ICU admission, and mortality with high sensitivity and specificity. Similarly, Brandwein et al. reported that pediatric patients admitted to the ICU with asthma exacerbations had higher pre- and post-treatment PVI values compared to those admitted to the ward or discharged [7]. Furthermore, our findings revealed that elevated PVI was independently associated with both ICU admission and mortality, suggesting that PVI measurement at triage may play a prognostic role in clinical decision-making.

Although post-treatment PVI decreased significantly after initial ED therapy, a fixed 2-hour measurement may reflect short-term physiological stabilization and may not capture later deterioration; in contrast, 30-day mortality is influenced by broader factors such as comorbidity burden and delayed complications [22]. According to our results, post-treatment PVI values were significantly lower than baseline values across all patient groups. Demir et al. reported similar findings in pediatric patients with asthma exacerbations, where post-treatment PVI values were significantly lower than baseline [6]. These results suggest that PVI may serve not only as a prognostic marker but also as a valuable parameter for monitoring treatment response in COPD exacerbations.

ABG analysis remains the cornerstone for objectively assessing disease severity and guiding early triage in COPD exacerbations [23]. The Rome classification identifies severe cases using hypercapnia (PaCO₂>45 mmHg) and acidosis (pH < 7.35) thresholds, underscoring the critical role of ABG in standardizing exacerbation severity [1, 24]. Current GOLD guidelines recommend continuous monitoring of oxygen saturation and serial ABG measurements in COPD exacerbations [1]. In our study, the negative correlation between PVI and oxygen saturation and pH, as well as its positive correlation with pCO₂, suggests that PVI may reflect a ventilation-perfusion mismatch and impaired gas exchange. These findings suggest that PVI may support clinical decision-making when invasive methods, such as ABG, are not feasible or time-consuming.

In addition to its correlation with arterial blood gas parameters, we formally evaluated the incremental prognostic value of PVI beyond conventional ABG variables. A baseline model including age, SaO₂, pH, pCO₂, and CRP demonstrated good discrimination for ICU admission; however, adding PVI significantly improved model performance, with a statistically significant increase in AUC. This finding indicates that PVI does not merely reflect the same physiological information as ABG parameters but also provides complementary prognostic information for early risk stratification. For 30-day mortality, although the improvement in AUC after adding PVI did not reach statistical significance, a numerically meaningful increase in discriminative performance was observed. This likely reflects limited statistical power due to the small number of mortality events and the multifactorial nature of longer-term outcomes. Nevertheless, these results suggest that PVI may contribute additional prognostic information beyond ABG parameters, particularly for short-term clinical deterioration requiring ICU admission.

Limitations

Several limitations of our study should be acknowledged. First, the proposed PVI cut-off values were derived and evaluated within the same cohort; therefore, given the single-center design and relatively small sample size, our findings may have limited generalizability and require external—preferably multicenter—validation before broad clinical adoption. Second, PVI measurements may be affected by peripheral perfusion abnormalities. Third, PVI measurements may be affected by the type and intensity of respiratory support (e.g., conventional oxygen therapy, high-flow nasal oxygen, or non-invasive ventilation) at the time of measurement. Although the respiratory support modality was recorded for each PVI assessment, residual confounding related to inter-patient variability in oxygen delivery and ventilatory settings may have influenced PVI values and limited the comparability of measurements across patients. Finally, the potential impact of cardiac comorbidities on PVI values could not be fully controlled. Larger multicenter studies are warranted to validate the reliability of PVI and to define its clinical role in COPD management better.

Conclusion

The findings of our study suggest that the PVI may serve as a valuable parameter for predicting exacerbation severity, hospital admission, and mortality in patients with COPD exacerbations. Due to its non-invasive, rapid, and objective advantages, PVI could be used in the ED as a complementary tool for triage, treatment decision-making, and patient monitoring. PVI should be interpreted as a complementary tool alongside clinical assessment and established laboratory parameters.

Abbreviations

ABG

Arterial Blood Gases

AUC

Area Under the Curve

CI

Confidence Intervals

COPD

Chronic Obstructive Pulmonary Disease

ED

Emergency Department

GOLD

Global Initiative for Obstructive Lung Disease

ICU

Intensive Care Unit

PP

Pulsus Paradoxus

PVI

Pleth Variability Index

ROC

Receiver Operating Characteristic

Authors’ contributions

Conceptualization: CKT, MY1, VVC,; Data curation: CKY, FC, EY, EAA, MO; Formal analysis: MY1, VVC, MG, MY2; Writing – original draft: CKT, MY1, VVC, HS; Writing – review and editing: CKT, MY1, MG, MY2, HS; Final approval of the version: CKT, MY1,VVC, MG, MO, HS, FC, EY, MY2, EAA.

Funding

The authors received no specific funding for this research.

Data availability

The datasets generated or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was designed as a single-center, prospective clinical investigation conducted in a tertiary care hospital. Approval was obtained from the local ethics committee (Samsun University Clinical Research Ethics Committee (SÜKAEK); protocol number: SÜKAEK/2023/17/15; date: 18.09.2023). Written informed consent to participate was obtained from all individual participants or their legal guardians prior to enrollment in the study.

Competing interests

The authors declare no competing interests.