P/PF ratio as a new criterion to classify the severity of acute respiratory distress syndromeABSTRACT
The Berlin criteria has classified ARDS since 2012, indicating its severity via P/F within 24 hours after diagnosis. It can identify severe patients but suits moderate and mild cases less precisely, necessitating the search for other tools that can accurately predict pulmonary impairment levels. A new evaluation uses PEEP levels for its definitions following P/PF. This cross-sectional study was carried out in ICUs at a public hospital in southern Brazil with patients aged over 18 years who were under mechanical ventilation and had been diagnosed with ARDS during hospitalization. The data that defined disease severity according to P/F and P/PF were compared on Stata®, version 12. The sample consisted of 85 patients, 91.7% of whom lied in the transmission window of COVID-19. The classification models showed differences (p<0.67) regarding P/F since they obtained moderate and severe results. Severe P/PF patients had a 54.9% average death rate, near the total average mortality rate in this study (55.29%), showing that P/FP better predicts severity in patients with ARDS due to its greater approximation of mortality percentages. However, furthers studies with larger samples and a greater number of classification variables may better predict ICU outcomes.
Keywords:
Acute Respiratory Distress Syndrome; Disease Severity Index; Invasive Mechanical Ventilation; Intensive Care Unit
RESUMO
A Síndrome do Desconforto Respiratório Agudo (SDRA) desde 2012 é classificada seguindo os critérios de Berlim, indicando sua gravidade por meio de relação P/F em 24 horas após diagnóstico, definindo com maestria pacientes graves, mas sendo menos adequada para pacientes moderados e leves, o que leva à busca por outras ferramentas que possam predizer os diferentes níveis de comprometimento pulmonar de forma acurada. Surge então uma nova forma de avaliação que utiliza os níveis variáveis de PEEP em momento diferente de definição, sendo demonstrada pela P/FP. Este estudo transversal foi realizado em Unidades de Terapia Intensiva (UTI) em um hospital público do Sul do Brasil com pacientes maiores de 18 anos que estavam sob ventilação mecânica e que foram diagnosticados com SDRA. Foram comparadas as gravidades da doença por meio da relação P/F e da P/FP, por meio do software Stata®, versão 12. A amostra contou com 85 pacientes, dentre os quais 91,7% estavam na janela de transmissão da covid-19. Os dois modelos de classificação diferiram entre si (p<0,67) na relação P/F, sendo moderada e grave na P/FP. Os pacientes graves pela P/FP tiveram uma média de óbitos de 54,9%, aproximando-se da média de mortalidade deste estudo, 55,29%, o que leva à conclusão de que a relação P/FP parece ser melhor preditora de gravidade dos pacientes com SDRA devido à maior aproximação de valores percentuais acerca da mortalidade. Porém, novos estudos com uma amostra maior e com um maior número de variáveis talvez possam ser mais assertivos em prever o desfecho na UTI.
Descritores:
Síndrome do Desconforto Respiratório Agudo; Índice de Gravidade de Doença; Ventilação Mecânica Invasiva; Unidade de Terapia Intensiva
RESUMEN
El síndrome de dificultad respiratoria aguda (SDRA) desde 2012 se clasifica según los criterios de Berlín que indica su gravedad mediante una relación P/F a las 24 horas del diagnóstico, definiendo magistralmente a los pacientes críticos, pero siendo menos adecuado para pacientes moderados y leves, lo cual lleva a la búsqueda de otras herramientas que puedan predecir con precisión los diferentes niveles de deterioro pulmonar. Entonces, surge una nueva herramienta de evaluación que utiliza los niveles variables de PEEP en un momento distinto de definición demostrada por P/FP. Este estudio transversal se realizó en unidades de cuidados intensivos (UCI) en un hospital público en el Sur de Brasil con pacientes mayores de 18 años que estaban en ventilación mecánica y que recibieron diagnóstico de SDRA. La gravedad de la enfermedad se comparó mediante la relación P/F y de P/FP con el uso del software Stata®, versión 12. La muestra estuvo conformada por 85 pacientes, de los cuales el 91,7% estaban en la ventana de transmisión de la covid-19. Los dos modelos de clasificación difirieron entre sí (p<0,67) en la relación P/F, y fueron moderados y severos en P/FP. Los pacientes graves por P/FP tuvieron una tasa de mortalidad media del 54,9% acercándose a la mortalidad media de este estudio, 55,29%, lo cual lleva a la conclusión de que la relación P/FP parece ser una mejor predictora de la gravedad de los pacientes con SDRA debido a la mayor aproximación de los valores porcentuales sobre la mortalidad. Sin embargo, nuevos estudios con una muestra más grande y un mayor número de variables pueden ser más asertivos en predecir el resultado en la UCI.
Palabras clave:
Síndrome de Dificultad Respiratoria Aguda; Índice de Severidad de la Enfermedad; Ventilación Mecánica Invasiva; Unidad de Cuidados Intensivos
INTRODUCTION
The Berlin criteria (2012) characterize acute respiratory distress syndrome (ARDS) by bilateral opacities on imaging, pulmonary edema partially explained by its cardiogenic origin, and hypoxemia1. Severity classification follows a positive end-expiratory pressure (PEEP) greater than or equal to 5cmH2O and a relationship between partial pressure of oxygen and fraction of inspired oxygen (PaO2/FiO2, P/F, in mmHg), deemed as mild (≤300), moderate (≤200), or severe (≤100)2.
Although severity classification based on the Berlin criteria may be useful for finding patients with severe ARDS, the instrument seems less suitable for distinguishing mild from moderate cases, which may impact prognoses, the risk for prolonged invasive mechanical ventilation, and mortality3.
Predicting the degree of pulmonary involvement requires evidence from treatment and measures that help to improve hospitalized patients and reduce mortality rates (estimated at 34.9% for mild cases, 40.3% for moderate ones, and 46.1% for severe ones4).
P/F ratio configures an easy-to-use everyday tool. However, it ignores varying PEEP levels and the nonlinear PaO2/FiO2 (directly linked to arterial oxygenation), which may fail to reflect lung injury intensity3),(4. Thus, classification models that more closely resemble the clinical picture can better reflect patients’ condition, especially regarding COVID-19 and its hyperinflammatory state, interstitial edema, hypoxemic respiratory failure, impaired pulmonary perfusion, and ventilation that often evolves to ARDS5.
Thus, studies have proposed a classification that uses PEEP [PaO2/(FiO2*PEEP) or P/FP (mmHg/cmH2O)] to predict the severity of respiratory failure on the third day of admission to an intensive care unit (ICU) by considering data from the second day after ARDS diagnosis rather than from the first one as in the Berlin criteria. It has the following thresholds: 60-40 (mild), 40-20 (moderate), and ≤20 (severe)3.
Thus, this study aimed to compare ARDS severity classification by using P/F and P/FP in patients and to evaluate which could serve as an improved criterion for predicting the severity and mortality of patients in ICUs.
METHODOLOGY
This cross-sectional study was carried out in adult ICUs at a public university hospital in southern Brazil from July to December 2021 with patients aged over 18 years who had received invasive mechanical ventilation and a diagnosis of ARDS during admission.
Patients who arrived intubated after the first day of ARDS diagnosis, those who remained on invasive mechanical ventilation for less than 48 hours or who died before 48 hours after ARDS diagnosis, and patients with previous heart failure were excluded from this study.
The first day was defined as the day in which the criteria for acute hypoxemic respiratory failure were diagnosed by the physicians on duty in the ICUs. If patients’ medical records showed no insufficiency but the individuals met the Berlin criteria, the patients were approached to confirm their diagnoses.
On the second day, the PaO2 and FiO2 values were recorded on the first day to define severity by P/F. On the third day, PaO2, FiO2, and the PEEP on the second day of ARDS diagnosis were collected to assess severity by P/FP. In addition to these data, other information was collected in a form to better characterize patients, including their age, gender, clinical diagnosis, days of sedation, laboratory and imaging tests, days of disease evolution after the first symptom, ventilatory parameters, measures adopted by the team, total days of orotracheal tube use, invasive mechanical ventilation time, length of hospital stay, and ICU outcome.
The data were tabulated and analyzed on Stata®, version 12. The data were first subjected to exploratory analysis and described by means and standard deviations. Then, the confirmatory analysis of the data was performed, assessing the association between the variables via analysis of variance and the Fisher’s and Bonferroni F tests. The Kruskal-Wallis test was used if the data failed to meet homoscedasticity assumptions. The chi-squared test6 was used to test differences between proportions, calculating incidence ratios and their 95% confidence intervals to quantify the found associations. Finally, interobserver agreement (which measures the degree of agreement beyond what would be expected by chance alone) was measured by the Kappa test. The level of statistical significance for the tests was set at 5%.
RESULTS
The final sample consisted of 85 patients, 56.47% of whom were men. Individuals had a mean age of 59 years. Among the diagnoses, 91.7% were positive for COVID-19; the remaining patients lied outside the transmission window of the virus. Mean ICU mortality (the outcome of this study) totaled 55.29%.
Measuring mean P/F values 24 hours after the ARDS diagnosis and those of P/FP at 48 hours obtained differing results regarding severity: moderate in P/F (161.28) and severe in P/FP (15.58). Mean PEEP at 24 hours totaled 12.07 and that at 48 hours, 11.81 (Table 1). These results showed no statistical significance (p<0.37).
The ARDS classification models disagreed:7 Kappa value: −0.016 (p<0.67).
Such lack of agreement emerges by considering that P/F only classified as severe 12 of the 73 patients P/FP deemed in such circumstance. The same occurred for moderate patients (six in P/F, 56 in P/FP).
To show the difference in deaths according to these classifications, Graph 1 depicts the average percentage of deaths according to P/F and P/FPs.
P/FP classified no patients as mild. Also, deaths followed an inverse order to the expected one. This study found more deaths in moderate cases than in severe ones despite the percentage of moderate patients only totaling 14.12%.
P/F found more deaths in mild cases than in moderate ones despite mild cases only totaling 18.82% of cases and moderate ones, 65.88%. Only 15.29% were classified as serious, showing a higher percentage of deaths (61.54%).
DISCUSSION
P/F can measure the severity of patients with ARDS, classifying cases (according to the Berlin criteria) as mild, moderate, and severe based on the values in the literature1. However, it ignores the currently used PEEP levels and the nonlinear PaO2/FiO2. Thus, it may fail to reflect lung injury intensity. An example of this would be a P/F that changes by incrementing PEEP values with the same FiO2, which would result in different PaO2 values8),(9.
New studies3),(8-10 have found a classification that uses PEEP, evinced by P/FP, and a change in the time of disease evolution for data collection: P/F uses data obtained in the first 24 hours of ARDS diagnosis and P/FP, from the 48 hours after diagnosis3.
Data collection at varying times (one classification 24 hours and another 48 hours after ARDS diagnoses) with another variable - in the case of P/FP (with PEEP) - could help predict the evolution and outcome of the disease and to outline treatment strategies. However, this study found no such results as PEEP averages at 24 and 48 hours resembled to each other. The found difference showed no statistical significance.
The classifications in this research differ significantly since P/FP evinced no mild cases and P/F found 18.2% of such cases. This difference between these models becomes more evident by considering that P/FP classified 85.88% of patients as severe and that P/F only deemed 15.29% of patients in the same situation.
As described, this may occur due to the lower specificity of P/F in differentiating mild and moderate patients5, differing from the mortality in the literature since, according to the severity classification and comparing relationships, 56.25% of individuals with mild cases according to P/F died, a rate quite unlike that of the literature, which predicts death in about 34.9% of mild cases.
Such result may stem from several factors due to the occasional need to increase PEEP values given higher fractions of oxygen. However, patients’ clinical conditions may prevent the use of such higher PEEP levels, including hemodynamic instability11) and injurious mechanical ventilation, which interfere with rescue measures due to the severity of the disease.
This brings to the discussion the Berlin criteria, which, unlike the evaluated ARDS severity categories, requires a minimum 5cmH2O PEEP. However, even if the evaluated classifications include minimum criteria, they ignore varying PEEP9.
Another factor that deserves attention is the 150 cutoff in P/F, which suggests rescue strategies despite the classified moderate severity. Some studies5),(12) have classified a P/F below this threshold as severe and values above 100 as very severe. This difference configures a weakness in defining the real severity of patients in ICUs and the ideal time to adopt strategies that reduce overall mortality.
Since the ARDS classifications disagree with each other, P/FP seems to better predict mortality than the Berlin criteria since it deemed 85.88% of patients as severe. Moreover, the 54.9% mean death rate closely resembles the mean mortality rate in this study (55.29%). This shows that P/FP obtained the closest results to the literature. Thus, it may have more accurately predicted mortality in this sample than P/F.
This necessitates studies with other variables in newer classification models (since using the PEEP at 48 hours was unable to obtain classifications that could accurately predict severity). They may find a new way of categorizing patients’ severity, affecting their prognoses and risk for prolonged mechanical ventilation, infections, and mortality in ICUs3.
The limitations of this study include its relatively small sample (n=85), hindering the obtention of levels of significance. The patients in this study could have received rescue maneuvers when classified within 48 hours of diagnosis. Patients with moderate P/FP may have shown a higher percentage of death than those in severe cases due to their prior status and a diagnosis of ARDS due to a disease the literature is yet to describe. Moreover, the discussion on how to address ARDS resulting from COVID-1913),(14 has mobilized arguments about the severity of the disease and the measures to be adopted.
CONCLUSION
P/F and P/FP classifications of ARDS severity differ from each other. P/FP seems to best predict mortality due to the greater approximation of its percentages to the mortality of patients admitted to ICUs and disease severity. Due to the high mortality rates of the disease, new classification models may include more variables to predict ARDS severity. Moreover, future studies with larger samples and a more assertive classification may propose therapeutic strategies to improve patients’ prognoses.
Data Availability:
The data underlying this study are available in the published article.
REFERENCES
-
1 Oliveira RP, Teixeira C, Rosa RG. Acute respiratory distress syndrome: how do patients fare after the intensive care unit? Rev Bras Ter Intensiva. 2019;31(4):555-60. doi: 10.5935/0103-507X.20190074
» https://doi.org/10.5935/0103-507X.20190074 -
2 Véras JB, Martinez BP, Gomes Neto M, Saquetto MB, Conceição CS, et al. Effects of prone position on patients with acute respiratory distress syndrome: a systematic review. Rev Pesqui Fisioter. 2019;9(1):129-38. doi: 10.17267/2238-2704rpf.v9i1.2175
» https://doi.org/10.17267/2238-2704rpf.v9i1.2175 -
3 Sayed M, Riaño D, Villar J. Novel criteria to classify ARDS severity using a machine learning approach. Crit Care. 2021;25:150. doi: 10.1186/s13054-021-03566-w
» https://doi.org/10.1186/s13054-021-03566-w -
4 Bellani G, Laffey JG, Pham T, Fan E, Brochard L, et al. Epidemiology, patterns of care, and mortality for patients with acute respiratory distress syndrome in Intensive Care Units in 50 countries. JAMA. 2016;315(8):788-800. doi: 10.1001/jama.2016.0291
» https://doi.org/10.1001/jama.2016.0291 -
5 Cunha MCA, Schardonga J, Righi NC, Lunardi AC, Sant’Anna GN, et al. Impact of prone positioning on patients with covid-19 and ARDS on invasive mechanical ventilation: a multicenter cohort study. J Bras Pneumol. 2022;48(2). doi: 10.36416/1806-3756/e20210374
» https://doi.org/10.36416/1806-3756/e20210374 - 6 Siegel S. Nonparametric statistics for the behavioral science. New York: Mcgraw-Hill; 1986.
-
7 Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159-74. doi: 10.2307/2529310
» https://doi.org/10.2307/2529310 -
8 Palanidurai S. P/FP Ratio (pao2/fio2 X Peep) X 10: a pilot approach to calculate the severity of oxygenation/ARDS with Peep. Am Thoracic Soc Int Conference Abstracts [Internet]. 2017 [cited 2026 Mar 4]. Available from: Available from: https://www.atsjournals.org/doi/epdf/10.1164/ajrccm-conference.2017.195.1_MeetingAbstracts.A3756?role=tab
» https://www.atsjournals.org/doi/epdf/10.1164/ajrccm-conference.2017.195.1_MeetingAbstracts.A3756?role=tab -
9 Palanidurai S, Phua J, Chan YH, Mukhopadadhyay A. P/FP: incorporation of PEEP into the PaO2/FiO2 ratio for prognostication and classification of acute respiratory distress syndrome. Ann Intens Care. 2021;11:124. doi: 10.1186/s13613-021-00908-3
» https://doi.org/10.1186/s13613-021-00908-3 -
10 Casado MS, Díaz MQ, Palacios D, Hortigüela V, Schulke CM, et al. Relationship between the alveolar-arterial oxygen gradientand PaO2/FiO2: introducing peep into the model. Med Intens. 2012;36(5):329-34. doi: 10.1016/j.medin.2011.10.007
» https://doi.org/10.1016/j.medin.2011.10.007 -
11 Beyer J, Meßmer K. Der Einfluß von PEEP-Beatmung auf Gesamthämodynamik und regionale Organdurchblutung. Klinische Wochenschrift, 1981;59(23):1289-95. doi: 10.1007/BF01711178
» https://doi.org/10.1007/BF01711178 -
12 Sehgal I, Agarwal R, Dhooria S, Prasad KT, Muthu V, et al. Risk stratification of acute respiratory distress syndrome using a PaO2. Lung India. 2020;37(6):473-9. doi: 10.4103/lungindia.lungindia_146_20
» https://doi.org/10.4103/lungindia.lungindia_146_20 -
13 Gil R, Bitar P, Deza C, Dreyse J, Florenzano M, et al. Cuadro clínico del covid-19. Rev Med Clín Las Condes. 2021;32(1):20-9. doi: 10.1016/j.rmclc.2020.11.004
» https://doi.org/10.1016/j.rmclc.2020.11.004 -
14 Robba C, Battaglini D, Ball L, Patroniti N, Loconte M, et al. Distinct phenotypes require distinct respiratory management strategies in severe covid-19. Respir Physiol Neurobiol. 2020;279. doi: 10.1016/j.resp.2020.103455
» https://doi.org/10.1016/j.resp.2020.103455
Edited by
-
Responsible editor:
Sônia LP Pacheco de Toledo
