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

Multicenter randomized controlled trial of the effects of inhaled nitric oxide on gas exchange in children with acute hypoxemic respiratory failure.

Dobyns EL, Cornfield DN, Anas NG, Fortenberry JD, Tasker RC, Lynch A, Liu P, Eells PL, Griebel J, Baier M, Kinsella JP, Abman SH.

J Pediatr 1999;134:406-412. [abstract]

Reviewed by Lisa Faberowski, MD, Children's Hospital, Boston

Review posted August 7, 2000

I. What is being studied?:

The study objective:

To determine whether early initiation of and prolonged therapy with inhaled NO can attenuate the progression of lung disease in pediatric patients with acute hypoxemic respiratory failure.

The study design:

Multicenter (7 centers), randomized, masked, placebo-controlled study

The patients included:

Any patient admitted to one of the seven participating center's ICU with acute hypoxic respiratory failure of sufficient severity to require intubation and mechanical ventilation. Acute hypoxic respiratory failure was defined as an OI > 15 measured by two serial blood gases six hours apart.

Although distinctly defined, this is a rather loose definition of AHRF as many people commonly conceive of it. This will include patients with multiple conditions, i.e., pneumonia, pulmonary edema, asthma, and not just patients with acute lung injury or acute respiratory distress syndrome. Though it could be argued that there are advantages to being inclusive and many of these conditions might benefit from iNO therapy on a theoretical basis, a mixture of pathophysiologic processes might tend to obscure true effects in a more homogeneous population.

The patients excluded:

Patients with congenital heart disease, cardiovascular surgery within the previous 14 days or a decision by the medical team and family not to provide full support.

The interventions compared:

Patients were randomized to receive inhaled nitric oxide or placebo for the first three days of the study. Nitric oxide was administered at 10 ppm to the patients randomized to the NO group. Individuals randomized to the placebo group continued have their ventilator and oxygen settings determined by the medical care team. Study treatment failure was defined as one of the following: 1) OI > 25 x 6 hours 2) OI >40 x 3 hours 3) progressive hemodynamic deterioration with systolic blood pressure < 50 mmHg or severe metabolic acidosis with pH < 7.2 that was unresponsive to colloid infusion and cardiotonic medications or methemoglobin level > 5% or nitrogen dioxide levels > 2 ppm.

Study treatment failure criteria were prospectively determined by the investigators to allow exit from the study and to provide for rescue therapy, even if the rescue therapy included NO.

After 72 hours, if failure criteria had not been met, in patients randomized to receive NO:

• NO was weaned to 3-5 ppm if the OI was >15
• A trial without NO if the OI was < 15,
  • If withdrawal caused a decrease in oxygenation (defined as needing an increase in FiO2 of at least 15%), NO was restarted at 3-5 ppm
• All NO patients were given a trial without NO daily until the withdrawal of NO was not associated with a significant decrease in oxygenation.
• If OI was maintained < 15 these patients exited the study.

After 72 hours, if failure criteria had not been met, in patients randomized to receive placebo, continued with placebo gas for 24 hours

• Patients who maintained an OI < 15 exited the study

The outcomes evaluated:

The outcome evaluated was whether fewer patients receiving NO would meet predefined and clinically relevant treatment failure criteria (defined above). Improvement or lack of improvement in oxygenation as determined by the oxygenation index in the 10 ppm NO group was compared to the placebo group during a 72 hour study period.

II. Are the results of the study valid?

1. Is there a strong, independent, consistent association between the surrogate end point and the clinical end point?

There is no known relationship between treatment failure at 72 hours (the surrogate endpoint) and mortality (the clinical end point) in acute respiratory failure. Nor is there any association that 72 hours signifies prolonged versus acute therapy. (1,2)

2. Is there evidence from randomized trials in other drug classes that improvement in the surrogate end point has consistently led to improvement in the target outcome?

No. Studies, most not randomized, involving the use of tolazamide, intravenous and inhaled prostacyclin have demonstrated similar evidence. (2,3) All of these pulmonary vasodilators show an acute improvement in oxygenation. However, none of these studies have demonstrated a difference in morbidity or mortality or a prolonged improvement in oxygenation.

3. Is there evidence from randomized trials in the same drug class that improvement in the surrogate end point has consistently led to improvement in the target outcome?

No. Most trials have not been randomized and all of the trials demonstrate an improvement in oxygenation in an acute manner. None of the previous investigations have demonstrated whether progressive deterioration in oxygenation can be improved in a prolonged manner. (1,2,3) Once again the definition of acute versus prolonged is not clearly defined.

IIa. Validity Questions for Therapy Articles:

Primary questions:

1. Was the assignment of patients to treatments randomized?

Yes, Treatment allocation was determined with a supply of randomization cards. However, the definition of randomization cards is not noted in the study design.

2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?

Was followup complete? and were patients analyzed in the groups to which they were randomized?

Not all patients were accounted for (or at least the calculated numbers are off). When examining treatment failures in table III the control group's ratio 7/55 is 49% not 54%, though they note that two patients were removed from the study. In the NO group, 23 /53 is a 43% failure rate, not the 56% reported.

It appears that all patients, except perhaps the two withdrawn for "clinical reasons" were followed up fully through the study course, though this is not made explicit. There is no data given beyond 72 hours and no data was reported on the crossovers.

There was crossover; however, from the failure criteria it is hard to determine which patients were the crossovers as demonstrated by the discrepancy in the number of failure patients. It does not appear that patients were permitted to cross over until they had been deemed a treatment failure.

Secondary questions:

3. Were patients, health workers, and study personnel "blind" to treatment?

Yes and no. Specific investigators at each site (usually a respiratory therapist and a research nurse) were aware of the assigned treatment. All NO tank regulators were covered to prevent unmasking of assignment. Methemoglobin levels were monitored at 6 and 24-hour intervals but the results were available only to the unblinded investigators. Monitoring of NO and nitrogen dioxide concentrations was also done by unblinded investigators.

Treatment allocation was not revealed to the masked investigators at the time of treatment failure.

4. Were the groups similar at the start of the trial?

The groups were similar as defined by patient age, primary diagnosis, pediatric risk of mortality score, mode of ventilation and median oxygenation index at the time of entry.

5. Aside from the experimental intervention, were the groups treated equally?

Clinical management was not monitored and compared. Although clinicians were allowed to manage the patients according to clinical judgement, they were blinded to the intervention making it unlikely that there was a different bias in the treatment of the groups. They did describe their approach to ventilator management in all patients but there was no information on the use of other aspects of therapy, i.e., nutrition, inotropes, etc.

III. What were the results?

1. How large, precise, and lasting was the treatment effect? (Effect should be large, precise, and lasting to consider a surrogate trial as possible basis for offering patients the intervention.)

The treatment effect may have been different if the patients had been enrolled earlier.

Treatment successes: The improvement in the PAO2/FIO2 ratio was greater in the NO group than the control group at both the 4- and 12- hour time points. See figure 2.

PAO2/FIO2	Control	NO	P
At enrollment	84+/-33	78+/- 30	NS
4 hours	98+/- 33	112+/- 48	0.016
12 hours	101 +/- 39	125+/-66	0.003

The results are expressed as a mean but the measure of difference is undetermined. It is important to note that in figure 2 that the authors graph the change in PAO2, and not the absolute values. Thus, the differences seem more impressive than if the absolute values had been graphed instead.

Furthermore, the "responders" are not well defined. One would assume that a responder was a non-failure patient, but this is an assumption. It is also not clear how much one's oxygenation had to improve to be a responder or if there were some patients who were neither responders nor failures.

The OI ratio decreased in both groups. (Figures 3 and 4 in the article). Once again the authors graph the change rather than absolute values, which makes the data again seem more impressive.

OI	Control	NO	P
At enrollment	30 +/- 15	35+/- 22	0.22
4 hours	27 +/- 9	25 +/- 19	0.014
12 hours	27+/- 8	26+/- 11	0.007

Figure 4: Rate of failure slowed by NO early in the study period as demonstrated by the KaplanÐMeier curve. The 72 hour decision point is unclear.

Subgroup analysis: In patients with an OI > 25 there appeared to be a difference in outcome at 72 hours.

1. In the control group a rapid decline in the number of responders occurred.
2. More than 50% met failure criteria before 10 hours of treatment
3. Whereas, 50% of the NO patients continued to respond until 40 hours
4. At 72 hours the NO group continued to have a greater proportion of responders compared to the control group. p = 0.046

Immunocompromised patients treated with NO had a more sustained response than those did in the control group (p = 0.033). There were only 9 immunocompromised patients in the iNO group, however, and one questions the effects of these small numbers on the validity of the conclusions.

Furthermore, without a statement that these subgroups were going to be evaluated a priori, this post hoc analysis can only be viewed as suggestive.

Treatment failures: The number of treatment failures was not different between groups as determined by OI criteria. The rate of failure appeared to be slowed by NO therapy, as shown in the Kaplan-Meier curve in figure 4. However, by 72 hours, there was no difference in the failure rate between the two groups.

Mortality: There was no difference between groups (overall 43%: 43.4% control vs. 42% iNO), but this was also not an endpoint.

In figure 6, the relative risk of treatment failure by 72 hours in the treatment group compared to the controls is 50%. However, this is a point estimate and without confidence intervals, it does not gives a range of where the point estimate of 0.5 might fall. Furthermore, it is a moot point since there was no difference in the groups with respect to the primary endpoint.

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

1. Can the results be applied to my patient care?

The patient population studied here is a mixture of typical multidisciplinary PICU patients with respiratory failure and is similar in most respects to many PICU patient populations. It is hard to argue for application of this technology for our patients, since there is no difference in treatment effect at 72 hours. Since it is suggested by subgroup analysis that there may be a potential benefit, studies in subgroups are necessary to determine which subgroup of patients may actually benefit from this method of treatment.

2. Were all clinically important outcomes considered?

This is the first study to address whether iNO has any effect beyond the first few hours. The clinically important outcome evaluated in this study was whether NO can attenuate progressive deterioration in a prolonged manner. One can conclude that the answer is no. However, mortality is the most important outcome and was not evaluated per se by this study.

3. Are the likely treatment benefits worth the potential harms and costs?

One can conclude that the treatment benefit was not worth the potential harms and costs. The treatment benefit hoped for ie. a prolonged improvement in oxygenation, was not demonstrated in this study. Mortality, though not an endpoint to the study, was not different between the groups.

References

1. Hoeper MM, Olschewski H, Ghofrani HA et al; A comparison of the acute hemodynamic effects of inhaled nitric oxide and aerosolized iloprost in primary pulmonary hypertension. J Am Coll Card. 1999. 35:176-182. [abstract]
2. Pappert D, Busch T, Gerlach H et al: Aerosolized prostacyclin versus inhaled nitric oxide in children with severe acute respiratory distress syndrome. Anesthesiology. 1995. 82: 1507-1511. [citation]
3. Bos AP, Tibboel D, Koot VC et al: Persistent pulmonary hypertension in high-risk congenital diaphragmatic hernia patients: incidence and vasodilator therapy. J Pediatr Surg. 1993. 28: 1463-5. [abstract]

Please note:

See an alternative review of this article here.

 

Comments

Back to the EB Journal Club Index