The Effect of On-Site Extracorporeal Membrane Oxygenation on the Therapy Choice and Outcomes of Neonates With Persistent Pulmonary Hypertension

The Effect of On-Site Extracorporeal Membrane Oxygenation on the Therapy Choice and Outcomes of Neonates With Persistent Pulmonary Hypertension* Paul F. Grim III, MD; Sandra K. Pope, MPH; Karl H . Karlson Jr ., MD, FCCP; and Bonnie ]. Taylor, MD

The effect of on-site extracorporeal membrane oxygenation (OS-ECMO) and selection criteria on the utilization rate of this technology is unknown. We retrospectively studied 55 neonates who were admitted to Arkansas Children's Hospital from 1985 to 1993. We compared the ECMO utilization, mortality, and morbidity rates for outborn neonates with moderate and severe persistent pulmonary hypertension (PPHN) before and after the establishment of an ECMO program with guidelines for its use at our institution. The rate of ECMO use was three times higher and the mortality rate was 13 times lower in the period after OS-ECMO compared with the period when ECMO was available only at other institutions. No differences were observed in the morbidity rates between the two periods. Physician decisions to initiate ECMO involved more than guidelines,

since 37% of the increased ECMO use was not associated with use of the guidelines. Possible reasons for noncompliance with the guidelines are discussed. Neonates who had received medical therapy only and who had an oxygenation index >30 and <40 had no mortality. Our findings suggest that the need for ECMO in this group of neonates is low. (Chest 1994; 106:1376-80)

Extracorporeal membrane oxygenation (ECMO) is a technique of extrathoracic cardiopulmonary bypass used in the rescue treatment of neonates with severe persistent pulmonary hypertension (PPHN). Some investigators have expressed concern about the increased use of this technology in these neonates. 1-4 In 1990, the Committee on Fetus and Newborn of the American Academy of Pediatrics recommended that each institution establish its own specific criteria for initiating ECM0. 5 Monitoring the compliance, utilization rates, and outcomes associated with such criteria may lead to a reduction in the inappropriate use of ECMO. Beginning in 1985, neonates from Arkansas Children's Hospital were transferred to out-of-state institutions for ECMO therapy. The decision to transfer the neonate was made when the attending neonatologist decided that the infant was unlikely to survive with medical therapy only. In July 1989, an ECMO program, including guidelines for its use, was established at our institution. We speculate that having on-site ECMO (OSECMO) may be associated with an increased utilization rate of this technology. The purpose of this

retrospective study was to compare the ECMO utilization, mortality, and major morbidity rates for neonates with moderate and severe PPHN before and after OS-ECMO. To determine the effect of guidelines on the utilization rate, physician compliance with guidelines in the period after OS-ECMO was examined.

*From the Department of Pediatrics, University of Arkansas for Medical Sciences and Arkansas Children's Hospital, Little Rock. Manuscript received May 27, 1993; revision accepted November 18. Reprint requests: Dr. Grim, 800 Marshall, Little Rock, AR 72202

1376

CT=computed tomography; ECMO=extracorporeal membrane oxygenation; HFJV=high-frequency jet ventilation; OI=oxygenation index; OS-ECMO=on-site extracorporeal membrane oxygenation; PPHN=persistent pulmonary hypertension

Keys words: extracorporeal membrane oxygenation; oxygenation index; persistent pulmonary hypertension

METHODS

The charts of all neonates who were admitted to the neonatal intensive care unit at Arkansas Children's Hospital from January 1985 to January 1993 were reviewed. The therapeutic option of venoarterial ECMO, in addition to medical therapy, was available to physicians treating neonates with PPHN who were admitted during this period. The period before OS-ECMO was from January 1985 to July 1989. The period after OS-ECMO was from July 1989 to January 1993. All neonates were born at other hospitals and transferred to our institution (outborn neonates). Each infant met the following criteria: birth weight 2::2 kg, gestational age 2::34 weeks, and postnatal age ~7 days 6 Each neonate also had at least one of the following diagnoses: aspiration syndrome, bacterial infection, pneumonia, or primary PPHN. Bacterial infection was documented by characteristic clinical and laboratory findings with confirmatory blood cultures and/ or with antigen studies. Neonates with congenital heart disease, lethal malformations, or coagulopathies were excluded, as well as those with pulmonary hemorrhage or central nervous system injury prior to admission at our institution (major intracranial hemorrhage, history of cardiac arrest, or seizures documented by EEG) 7 In addition, all neonates with respiratory distress syndrome and secondary PPHN were excluded because our participation in the double-blinded

Extracorporeal Membrane Oxygenation in Neonates With Pulmonary Hypertension (Grim eta/)

Table !-Characteristics of Neonates With Moderate and SeverePPHN* Severe (0!2:40)

Moderate (0!2:30<40)

Variables Birth weight, g Gestational age, wk Mode of delivery, vaginal Apgar .:::;3 at 5 min Respiratory diagnosis Primary PPHN Aspiration Infection/ pneumonia Age at time of hospital admission, h 01 at admission 012:40 at admission

Before OS-ECMO (n=7)

After OS-ECMO (n=12)

Before OS-ECMO (n=16)

After OS-ECMOt (n=20)

3,115±373 40.1±2.1 4 (57) 2 (29)

3,360±765 39.3±2.4 7 (58) 1 (8)

3,393±623 39.3±1.8 10 (63) 1 (6)

3,612±763 40.0±2.1 10 (50) 0

2 (29) 5 (71) 0 17.3±15.6 29.9±3.0 0

2 (17) 9 (75) 1 (8) 18.6± 18.8 22.8±9.0 0

1 (6) 10 (63) 5 (31) 13.0±7.2 48.1±25.2 10 (63)

6 (30) 10 (50) 4 (20) 19.5±14.8 35.3±20.1 8 (40)

*Oxygenation index (01) values were based on three to five arterial blood gas determinations. Continuous data represent mean± SD; categorical data represent number and (percent) neonates. No statistically significant differences were observed between the before and after OS-ECMO groups with moderate or severe PPHN. tTwo neonates with severe PPHN were treated with ECMO before having three 01 values 2:40. However, both neonates did have several OI values 2:40 at the referring hospital. prospective surfactant replacement therapy trials from 1988 to 1990 would confound the outcome of some of these patients. Neonates who received ECMO from our mobile ECMO team before arriving at our institution were also excluded. All infants had at least one postductal Pa02 .:::;70 while receiving assisted .ventilation with Flo2=l.OO and ventilator rate 2:60 for at least 6 h. 1 They also had an oxygenation index (OI) of 30 or greater. This index was calculated by dividing the product of Flo2 (X100) and the mean airway pressure by the postductal Pa02. 7 The 01 was calculated using three to five of the postductal arterial blood gas determinations. These determinations were separated by at least 30 to 60 min. The 01 values from the referring hospital and our hospital were calculated. In 1989, guidelines for initiating ECMO at our institution were established based on a 4-year retrospective study. The following guidelines were recommended: acute deterioration (Pa02.::;4o or pH.::;7.15 for 2 h) or an OI 2:40. 8 To determine if phy~ician compliance with these guidelines affected the use of ECMO, documentation of the 01 or acute deterioration and treatment associated with the guidelines were noted. To control for differences that may have been due to the severity of PPHN, we stratified our analyses according to the neonate's highest level of respiratory acuity: moderate PPHN

Table 2-ECMO Use in Neonates With Moderate and Severe PPHN Before and After On-Site ECMO With Guidelines* Moderate (0!2:30<40)

(01,2:30<40) and severe PPHN (012:40). The demographic, diagnostic, therapeutic, mortality, and morbidity variables of the infants were examined. Continuous data were analyzed using the Student's t test. Fisher's Exact Test or x2 test were utilized to examine categorica,l data. A probability value of .:::;0.05 was considered statistically significant. RESULTS

A total of 55 neonates were included in this study. None of the neonates with moderate or severe PPHN qualified for ECMO by acute deterioration criterion. Neonates with moderate or severe PPHN from the before and after OS-ECMO periods did not differ according to demographic, diagnostic, or respiratory acuity variables (Table 1). Five neonates in the after OS-ECMO group received high-frequency jet ventilation (HFJV). Two of these five were later treated with ECMO. Five of the 23 neonates (22%) treated in the period before OS-ECMO received ECMO vs 24 of the 32 (75%) treated after OS-ECMO (Table 2). Mortality data according to therapy is shown only for neonates Table 3-Mortality Data for Neonates With Severe PPHN

Severe (0!2:40)

Before After Before After Therapy OS-ECMO OS-ECMO OS-ECMO OS-ECMO Totals Medical ECMO Totals

7 0 7

5 7 12

11 5 16

3 17 20

26 29

55

*The OI values were calculated while the neonates received conventional mechanical ventilation. Criteria for initiating ECMO at our institution were three OI values 2:40 from three to five successive blood gas determinations and acute deterioration. ECMO use increased after OS-ECMO in both neonates with moderate and severe PPHN compared with before OS-ECMO with p values of <0.02 and 0.005, respectively.

Therapy

Before OS-ECMO, No.(%) n=16t

After OS-ECMO, No.(%) n=20t

Totals, No.(%) n=36

Medical ECMO Totals*

7/ 11 (64) 2/ 5 (40) 9/ 16 (56)

0/3 (0) 1/ 17 (6) 1/ 20 (5)

7/14 (50) 3/22 (14) 10/36 (28)

*Significantly fewer deaths occurred in the after OS-ECMO than before OS-ECMO period (p.::;0.002). fNeonates with 012:40 at hospital admission (10/16); after admission (6/16). tNeonates with 012:40 at hospital admission (8/20); after admission (12/ 20). CHEST /106/5/ NOVEMBER, 994

1377

Table 4-Morbidity Data for Neonates With PPHN* Moderate PPHN, No. (%)

Severe PPHI'\, No. (%)

( 0! ~30<40 )

( 0! ~40 )

Variable

Medical Therapy (n=l2 )

ECMO Therapy (n=7)

Medical Therapy (n=l4)

ECMO Therapy (n=22)

Chronic lung disease, f Fioz >0.21 at 1 mo Intracranial abnormalityt EEG abnorm ality§ Feeding problems in survivor~!

1/ 12 (8) 0/ 9 (0) 0/ 0 (0) 1/ 12 (8)

1/ 7 (14) 2/ 7 (29) 0/ 2 (0) 2/ 7 (29)

1/7 (14) 0/ 11 (O) 1/ 1 (100) 0/ 7 (O)

3/20 (15) 6/ 22 (27) 6/8 (75 ) 6/ 19 (32)

*Results s hown as number of neonates with positive test divided by number with diagnostic test. tOne neonate who had severe PPHN and was treated with ECMO developed chronic lung disease and died after 1 month of age. t Hemorrhagic and nonhemorrhagic abnormalities noted on magnetic resonance imaging, CT, ultrasound, o r at autopsy. Neonates who had moderate and severe PPHN and received ECMO therapy had significantly more intracranial abnormalities than those receiving medical therapy (p=0.03 ). §Sharp-wave activity, low voltage, or decreased reactivity. lrrube feedings for over 2 weeks or gastroesophageal reflu x documented by upper gastrointestinal series or technetium scan.

with severe PPHN since there were no deaths in the group with moderate PPHN (Table 3). Reduced mortality rates were noted in the period after OSECMO regardless of the therapy. A significantly lower mortality rate was noted in the neonates from the after compared with the before OS-ECMO period (1 / 20 vs 9/ 16; p<0.002). Further definition of this improvement in the total mortality rate was accomplished by noting the mortality rate of these neonates with severe PPHN according to their acuity at the time of hospital admission (Table 3). The mortality rate in neonates who were admitted with 01 >40 after OS-ECMO was 1 of 8 (12.5%) compared with 6 of lO (60 %) in those who were admitted from the period before OSECMO. This difference in mortality rate was not statistically significant (p=0.07) . Neonates who developed 01 >40 after hospital admission and were from the period after OS-ECMO had significantly reduced mortality when compared with those from the period before OS-ECMO (0/ 12 vs 3/ 6; p=0.03). Morbidity data in neonates treated before and after OS-ECMO were compared according to respiratory acuity. No significant differences were noted. The results are not shown . We also compared morbidity data in neonates with moderate and severe PPHN according to therapy (Table 4) . The only significant difference was the increased incidence of intracranial abnormalities in neonates treated with ECMO compared with those treated medically . Most of the neonates who were treated medically had neuroimaging by ultrasonography only. In contrast, most who were treated with ECMO had intracranial evaluations by ultrasound during ECMO and by computed tomography or magnetic resonance imaging after ECMO. Eight neonates treated with ECMO had intracranial abnormalities. One of these neonates died. Hemorrhagic lesions were noted in five neonates, cerebral infarcts in two, and cerebral atrophy 1378

in one. The distribution of the lesions showed no predominant laterality. Data regarding physician compliance with the institution-specific guideline is shown in Table 5. Five of the seven neonates who had moderate PPHN and were treated with ECMO had at least one 01 >40. Three of the seven had either two consecutive Ols >40 within 1 h or rapid deterioration and were placed on ECMO. Seven neonates had moderate PPHN before OS-ECMO. Five of these neonates had a single OI >40 and four had two Ois >40. All survived. DISCUSSION

We have shown that for neonates with PPHN and 01>30, the rate of ECMO use was 31fz times higher and the mortality rate was 13 times lower in the period after the establishment OS-ECMO program when compared with the period when ECMO was available only at other institutions. The neonates whose mortality rate was reduced the most after Table 5-Physician Compliance With Guideline for Initiating ECMO* Type o f Therapy Patient Group

No. of Documentation Neonates of Guideline

Moderate PPHNt

12

Severe PPH Nl

20

Totals

32

Yes No Yes No

~

Medical

ECMO

2 3 0 3 8

5 2 15 2§ 24

*Compliance has two parts: (1)documentation of the guideline that is any reference to or calculation of our guideline (oxygenation index or OI ) in chart and (2) therapeutic compliance wit h the guideline. fThese neonates did not have three of five OI values ~ 40 and therefore did not qualify for ECMO. f These neonates did qualify for ECMO. §One of these neonates died.

Extracorporeal Membrane Oxygenation in Neonates With Pulmonary Hypertension (Grim et alj

OS-ECMO were those who developed severe PPHN after admission to our institution. Physicians documented and acted in accordance with the 01 criterion in 15 of 24 patients (63%) who had moderate and severe PPHN and who were treated with ECMO. The marked reduction in the mortality rates in the period after OS-ECMO compared with the period before OS-ECMO may be attributable to two factors . First, 63% of the neonates who had an 01 >40 and were from the period before OS-ECMO were admitted to the hospital with severe PPHN. Their acuity may have been considered too great for safe transfer to a distant medical center for ECMO and thus may have contributed to the observed medical mortality (64%). Likewise, if they were transferred, their acuity at the time of admission to our institution may have contributed to the observed "off-site ECMO" mortality (40%). This mortality rate was high when compared with neonates who had similar clinical characteristics and who were treated with ECMO from 1985 to 1989. 9 The increased mortality of neonates treated with "off-site" compared with OS-ECMO would support the suggestion of hidden mortality associated with transport described by Boedy et al. 10 Secondly, ECMO was initiated at a lower level of acuity in the period after OS-ECMO compared with the period before OS-ECMO. This practice and improved ECMO technology may have contributed to the improved mortality in this period. Neonates with the characteristics defined in this study and with an 01 >30<40 had a low mortality risk. Over the past 7% years, we have treated 12 such neonates with medical therapy only and no deaths occurred (three had Ols >35<40). Only one of these 12 neonates received HFJV . This finding suggests that the need for ECMO in neonates with an 01 >30<40 is low . This study further shows that neonates treated with ECMO had significantly more intracranial abnormalities than those treated medically. There may be several reasons for this finding . First, the ligation of the right common carotid artery and jugular vein and systemic anticoagulation required with ECMO may predispose certain neonates to cerebrovascular injury.11,12 Second, the increased survival rate in the ECMO-treated infants compared with the survival rate of those who were medically treated may have increased the incidence of intracranial abnormalities in the ECMO-treated group. This "survival bias" may occur because the survivors have a longer period for the development and detection of brain lesions than nonsurvivors. Third, some intracranial abnormalities in medically treated infants may have been missed since neuroimaging was not done in all of them . Finally, the increased incidence may reflect the more sensitive neuroimaging methods used in

ECMO survivors compared with the methods used in neonates treated medically.l 2 •13 Physician noncompliance with our guidelines in neonates who had moderate PPHN and who were from the period after OS-ECMO suggests that factors other than the respiratory criteria affected the decision to use ECMO. One of these factors may be the physician's awareness of the limitations of the guidelines. The neonates who were from the period after OS-ECMO and who were treated with ECMO even though not meeting the specified guidelines were more unstable and had higher acuity (five of seven had 01 >35<40) than those treated medically. Treatment of these neonates with ECMO suggests that physicians assumed that the conditions of these more unstable neonates would deteriorate and subsequently they would require ECMO. However, the survival of all members of a similar unstable group who were from the period before OS-ECMO and who received only medical therapy casts some doubt on this assumption. These observations confirm the need for a better estimation of the mortality rate of neonates with moderate PPHN. This may prevent early, and in some cases unnecessary, utilization of ECMO in these patients. Another factor that may explain the observed physician noncompliance in these patients is the expectation of improved outcome. Our physicians may have initiated ECMO before the infants met our guidelines in order to meet the expectations of the staff , referring physicians, and families regarding reduced mortality and morbidity in neonates with PPHN when treated with ECMO. Our physicians also realize that the referring physicians may change referral patterns or transfer criteria when an adverse outcome occurs in "their" patients while receiving medical therapy and awaiting qualification for ECMO by guidelines. In summary, we have shown that the establishment of an ECMO program at our institution was associated with an increased ECMO utilization rate and a marked reduction of the mortality rate of outborn neonates with moderate and severe PPHN when compared with a period when ECMO was available only at other institutions. The reduction in the mortality rate of neonates with severe PPHN from the period after OS-ECMO compared with those before OS-ECMO was attributed to increased ECMO use in the later period. Neonates from this study who had moderate PPHN and who were treated medically had no mortality. This finding suggests that the need for ECMO in these infants is low. Finally, the physician decision to initiate ECMO involved more than the use of guidelines since approximately 37% of this increased ECMO use was not associated with the use of the guidelines. CHEST I 106 I 5 I NOVEMBER, 994

1379

REFERENCES

2 3 4

5 6

7

Cole CH, Jillson E, Kessler D. ECMO: regional evaluation of need and applicability of selection criteria. Am J Dis Child 1988; 142:1320-24 Dworetz AR, Moya FR, Sabo B, Gladstone I, Gross I. Survival of infants with persistent pulmonary hypertension without extracorporeal membrane oxygenation. Pediatrics 1989; 84:1-6 Short BL, Brans YW. Extracorporeal membrane oxygenation: pro and con. Curr Opin Pediatr 1990; 2:308-14 Payne NR, Kriesmer P, Mammel M, Meyer C. Comparison of six ECMO selection criteria and analysis of factors influencing their accuracy. Pediatr Pulmonol199l ; 11:223-32 American Academy of Pediatrics Committee on Fetus and Newborn Pediatrics. Recommendations on extracorporeal membrane oxygenation. Pediatrics 1990; 85:618-19 Bartlett RH, ToomasianJ, Roloff D, Gazzaniga AB, Corwin AG, Rucker R. Extracorporeal membrane oxygenation (ECMO) in neonatal respiratory failure-100 cases. Ann Surg 1986; 204: 236-45 Bartlett RH, Roloff D, Cornell RG, Andrews AF, Dillon PW, Zwishenberger JB. Extracorporeal circulation in neonatal respiratory failure: a prospective randomized study. Pediatrics 1985; 76:479-87

1380

8 Grim PF, Taylor BJ, Kirby R. Oxygenation index, acute deter~orati~n and mortality in neonates with pulmonary hypertenSIOn without respiratory distress syndrome [abstract). Clin Res 1990; 38:63A 9 Stolar CJ, Snedecor SM, Bartlett RH. Extracorporeal membrane oxygenation and neonatal respiratory failure: from the extracorporeal life support organization. J Pediatr Surg 1991; 26:563-71 10 Boedy RF, Howell CG, Kanto WP Jr. Hidden mortality rate associated with extracorporeal membrane oxygenation. J Pediatr 1990; 117:462-64 11 Cilley RE, Zwischenberger JB, Andrews AF, Bowerman RA, Roloff DW, Bartlett RH. Intracranial hemorrhage during extracorporeal membrane oxygenation in neonates. Pediatrics 1986; 78:699-704 12 Taylor GA, Fitz CR, Miller MK, Garin DB, Catena LM, Short BL. Intracranial abnormalities in infants treated with extracorporeal membrane oxygenation: imaging with US and CT. Radiology 1987; 165:675-78 13 Wiznitzer M, Masaryk TJ, Lewin J, Walsh M, Stork E. Parenchymal and vascular magnetic resonance imaging of the brain after extracorporeal membrane oxygenation. Am J Dis Child 1990; 144:1323-26

Extracorporeal Membrane Oxygenation in Neonates With Pulmonary Hypertension (Grim et al)