Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Elevated Plasma Surfactant Protein-B Predicts Development of Acute Respiratory Distress Syndrome in Patients with Acute Respiratory Failure.

Bersten AD, Hunt T, Nicholas TE, Doyle IR.

Am J Respir Crit Care Med 2001;164 648-652. [abstract; full-text for subscribers]

Reviewed by Sally Huntoon Vitali, MD, Children's Hospital Boston

Review posted May 22, 2003

I. What is being studied?:

The study objective:

To determine whether plasma levels of surfactant protein-A and B (SP-A and SP-B) are predictive of development of ARDS in patients admitted to an ICU for mechanical ventilation.

The study design:

Prospective, nested case-control study with blinded quantification of SP-A and SP-B levels.

The study included 54 consecutive, consenting patients admitted to one critical care unit who:

• required respiratory support with either noninvasive or intubated ventilation
• had an identifiable risk factor for ARDS as defined by Gregory and coworkers (1) (pneumonia, aspiration of gastric contents, trauma, massive transfusion, sepsis, or pancreatitis) or "other risk factors" of preeclampsia, lung contusion, and pulmonary embolism
• had plasma samples and matching physiologic data for calculation of the lung injury score collected within 8 hours
• had a lung injury score of less than 2.5 (and therefore did not yet meet criteria for ARDS)

The study excluded patients who:

• had left-ventricular failure as a cause for their respiratory failure

II. Are the results in the study valid?

Primary questions:

1. Was there a representative and well-defined sample of patients at a similar point in the course of the disease?

The individuals in the study were a well-defined population of patients, but it is not clear how representative the group is of the general population. Presumably, all patients who eventually develop ARDS would present to an intensive care unit during the evolution of their disease, and if this is a general medical and surgical intensive care unit that serves a diverse group of patients, then it should be a representative group. The authors provide a breakdown of the number of patients with different ARDS risk factors (diagnoses) in table 1, and this set of patients appears to be similar to the usual adult ICU population.

Although the diagnoses are distributed in a usual pattern, the patients evaluated in the study may have differed at study enrollment by underlying co-morbidities. The authors do not mention whether, for example, many of the patients with direct lung injury causing their need for mechanical ventilation also had COPD or lung cancer. If this were the case, then evaluating the distribution of prognostic test results and outcome variables in patients with different sets of co-morbidities might explain the differences that the authors found between the patients who developed and didn't develop ARDS. A description of co-morbidities would help ensure representativeness of the patient group.

The patients evaluated were generally in a similar point in their disease course. Obviously, some patients progress rapidly to ARDS and others more slowly, and it is possible that some patients may have been ill requiring oxygen for many days prior to requiring mechanical ventilatory support, while others progressed quite rapidly from wellness to intubation. Many definitions of ARDS, including the 1994 American-European Consensus Conference definition, note that ARDS is a disease of acute onset. Reporting the duration of hospitalization prior to enrollment and potentially stratifying results based on this criteria (long duration vs. short) would be helpful in determining whether this factor was important in determining serum SP-B levels or development of ARDS. Nevertheless, patients were enrolled in the study at a specific point when they developed the need for mechanical ventilation but had not yet progressed to a high lung injury score, and one could argue that this is a well-defined and specific time in the course of these patients' illnesses.

It is not clear how many patients were not enrolled in the study because they were not evaluated early enough in their disease course and already had a LIS of > 2.5 at the time of evaluation. It is also not clear how many patients were not enrolled because they refused consent or because blood was not collected promptly enough. The number and characteristics of the excluded patients would have helped demonstrate representativeness.

2. Was follow-up sufficiently long and complete?

Yes. The study followed patients to hospital discharge, which was sufficiently long enough to determine whether Surfactant Protein-B predicted development of ARDS. The analysis of whether Surfactant Protein-B predicted hospital mortality was sufficient, since patients who die from ARDS would usually do so during the same admission in which they were diagnosed.

3. Were outcome variables clearly and objectively defined?

Yes. Outcomes evaluated in this study were:

• Development of ARDS as defined by:
• Bilateral diffuse lung infiltrates on the chest radiograph
• Lung injury score > 2.5
• Ventilator-free days: Number of days up to day 28 during which the patient was free of invasive or noninvasive ventilator support.
• Hospital mortality
• Plasma creatinine

It was important that the definition of ARDS be clear and as objective as possible, as there is no gold standard test for making the diagnosis. The lung injury score was used by the investigators because it included the amount of PEEP in the evaluation. They note that 1994 American-European Consensus Conference "discarded" the LIS for a definition that focused on PaO2/FiO2 ratios, but that the GOCA score embraced by the 1998 AECC is similar to the LIS. They note that a study by Meade and others (2) found that in general, the two scoring systems (1994 AECC and LIS) could be used interchangeably but that there were some patients who met criteria by one system and not the other. As noted above, ten of the patients in this study would have already developed ARDS at the time of enrollment by the 1994 AECC definition. Of note, when Meade's group compared patients who only met criteria for ARDS by LIS and those who only met criteria by 1994 AECC, mortality was lower in the AECC group versus the LIS group, although this was not statistically significant because of the size of these groups. Both the LIS and AECC scoring systems have been used throughout the ARDS literature. For example, in the 1997-98 studies of low tidal volume/protective lung strategy for management of ARDS, Brochard and others (3) and Amato and others (4) used the LIS scoring system, while Brower and others (5) used the 1994 AECC definition.

The authors note that if the 1994 American-European Consensus Conference definition of ARDS (namely, PaO2/FiO2 ratio) had been used in their study, 10 of their 54 patients would have already had ARDS at study enrollment. They do note that both with and without these patients in their cohort, plasma SP-B was significantly higher in patients who went on to develop ARDS. They clarify that five of these patients had direct lung injury and 5 had indirect lung injury, but it is not clear whether these patients went on to develop ARDS by the study criteria or not. They point out that the PaO2/FiO2 ratios of all patients at study entry were relatively low and that these were not predictive of ARDS development, so these ten patients may not have ever developed a high enough LIS to be defined as ARDS in the study. Interestingly, even on days 1-3, PaO2/FiO2 ratios were not significantly different among those who had and had not developed ARDS. The authors point out that this further validates their use of the LIS instead of the 1994 AECC criteria.

III. What are the results?

1. What is(are) the prediction tool(s)?

The study found no significant difference in day 0 SP-A or LIS between developers and non-developers of ARDS. Although no significant difference in day 0 SP-B was found for patients with indirect lung injury who went on to develop ARDS, there was a significant difference in the group with direct lung injury. ROC curves were analyzed and AUC's that were significantly different from the line of no information were found for SP-B in all patients (0.77 with 95% confidence interval 0.63-0.90) and in those with direct lung injury (0.87 with 95% confidence interval 0.72-1.02). For patients requiring mechanical ventilation (invasive or non-invasive) because of a direct lung insult, the authors describe an optimal cutoff point for SP-B on day 0 in as being 4,994 ng/mL. ROC predicts that levels at or above this value should be 85% sensitive and 78% specific for that a patient will develop ARDS.

2. How well does the model characterize patients into different levels of risk?

The fact that neither SP-A, SP-B, or LIS on day 0 were able to predict ARDS development in the indirect lung injury group indicates the tests are not useful for that group of patients. For a seemingly homogeneous group of patients with direct lung injury, the study provides only a cut-off value for SP-B above which ARDS development would be more likely and below which it would be less likely.

3. How confident are you in the estimates of risk?

Only somewhat confident. At best, in patients with a direct lung injury, a SP-B concentration of > 4,994 ng/mL predicted ARDS with a sensitivity on 85% (95% confidence interval of 40-97%), a specificity of 78% (95% confidence interval 55-98%), a positive predictive value of 85% (95% confidence interval 55-98%), and a negative predictive value of 78% (95% confidence interval 40-97%). These confidence intervals are broad and reflect that the direct lung injury group was small and variability amongst the values was large. Looking back at figure 2, we see that the appropriate nonparametric Mann-Whitney U test and showed that the difference in day 0 SP-B between developers and non-developers of ARDS with direct lung injury was significant at the p=0.001 level. The 25th-75th percentiles (interquartile ranges) for day 0 SP-B do not overlap for the direct lung injury group (see figure 2).

IV. Will the results help me in caring for my patients?

1. Does the tool maintain its prediction power in a new sample of patients?

The authors do not test the 4,994 ng/mL cutoff point in a new set of patients.

2. Are your patients similar to those patients used in deriving the diagnostic tool?

No. The study patients were adults with a wide variety of underlying illnesses and causes of ALI. Some of these causes were presumably similar in pathophysiology to their corresponding condition in the pediatric population, but others may have different means of affecting lung damage in the pediatric age group. These patients had an average age of 67 years and their co-morbidities were not elucidated in the article. Pediatric patients are likely to have fewer chronic conditions, and those chronic conditions that some pediatric patients have are very different from adult conditions.

The percentages of pediatric patients requiring mechanical ventilatory support who progress to develop ARDS has not been reported and may be different from the adult population.

Further study is needed to determine whether plasma SP-B is equally affected by diffuse alveolar damage in the different age groups of the pediatric population before the test can be applied universally to pediatric patients.

3. Will the results lead directly to selecting or avoiding therapy?

No. SP-B measurement is not readily available. More importantly, there are no therapies available that could be selected for early treatment of ARDS In the future, if preventative measures or therapies exist that have inherent risks, and SP-B could be measured quickly and easily, then this test has the potential for helping to define the patients who would most directly benefit from a therapy. At this point, patients with acute lung injury are managed in a similar manner regardless of their presumed likelihood of going on to develop ARDS.

4. Are the results useful for reassuring or counseling patients?

No, in terms of the use of SP-B levels, for the aforementioned reasons. However, the results are interesting in that they show that PaO2/FiO2 ratios and lung injury scores at the time of initiation of mechanical ventilation are not useful in predicting which patients will go on to develop ARDS in the subsequent days. ARDS is a possibility for almost all patients who present to the ICU for mechanical ventilation and initial discussions with families can include this information.

Appendix A: Lung Injury Score (LIS) of Murray (6)

Chest Radiograph Score	No Consolidation	0
	One quadrant	1
	Two quadrants	2
	Three quadrants	3
	Four quadrants	4
Hypoxemia Score (PaO2/FiO2)	≥300	0
	225-299	1
	175-224	2
	100-174	3
	<100	4
Respiratory Compliance	>0.04 cc/cmH20/ccFRC	0
	0.02-0.03	1
	0.02-0.04	2
	0.01-0.02	3
	<0.01	4
PEEP	≤5 cm H20	0
	6-8	1
	9-11	2
	12-14	3
	≥15	4

The final score is obtained by dividing the sum of scores by the number of components used.

Appendix B: 1994 AECC definition of ARDS (7)

• Acute onset of respiratory failure
• PaO2/FiO2 ratio less than 200 regardless of PEEP
• Diffuse bilateral infiltrates on CXR
• PCWP < 18 cm H20 or no evidence of left atrial hypertension.

References

1. Gregory TJ, Longmore WJ, Moxley MA, Whitsett JA, Reed CR, Fowler AA, Hudson LKD, Maunder RJ, Crim C, Hyers TM. Surfactant chemical composition and biophysical activity in acute respiratory failure. J Clin Invest 1991; 88:1976-1981. [abstract]
2. Meade MO, Guyatt GH, Cook RJ, Groll R, Kachura JR, Wigg M, Cook DJ, Slutsky AS, Stewart TE. Agreement between alternative classifications of acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163: 490-493. [abstract]
3. Brochard L, Roudot-Thoraval F, Roupie E, Delclaux C, Chastre J, Fernandez-Mondejar E, Clementi E, Mancebo J, Factor P, Matamis D, Ranieri M, Blanch L, Rodi G, Mentee H, Dreyfuss D, Ferrer M, Brun-Buisson C, Tobin M, Lemaire F. Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. Am J Respir Crit Care Med 1998; 158:1831-1838. [abstract]
4. Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Olivera R, Takagai TY, Carvalho CR. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. New Engl J Med 1998; 338:347-354. [abstract]
5. Brower R, Stanholtz C, Shade D, Fessler H, White P, Wiener C, Teeter J, Almog Y, Dodd-O J, Piatadosi S. Randomized trial of small tidal volume ventilation (STV) in ARDS. Am J Respir Crit Care Med 1997; 155: A93.
6. Murray JF, Matthay MA, Luce JM. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138: 720-723. [abstract]
7. Bernard GR, Artigas A, Brigham KL, Carlet J, Flake K, Hudson L, Lamy M, Legall JR, Morris A, Spragg R, and the Consensus Committee. The American-European consensus conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149: 818-824. [abstract]

 

Comments

Back to the EB Journal Club Index