Gastrointestinal perforation in infants

Gastrointestinal Perforation in Infants Anthony P. Borzotta, MD, Cleveland, Ohio and Diller B. Groff, MD, Louisville, Kentucky

P e r f o r a t i o n of the gastrointestinal t r a c t in infants and children is associated with m o r t a l i t y rates of 42 to 71 p e r c e n t [1-4]. In 1985, Bell [5] described imp r o v e m e n t s in overall survival, reflecting a d v a n c e s in ventilatory and metabolic support, d e v e l o p m e n t of centers devoted to neonatal intensive care, and an increased capability and willingness to operate on critically-ill newborns. Most pediatric centers t r e a t only three to four cases annually, providing few o p p o r t u n i t i e s to review this condition. T h e p r e s e n t s t u d y is an analysis of our experience with perforation of the gastrointestinal t r a c t in infants. Statistical analysis was p e r f o r m e d utilizing the chi-square m e t h o d with Yates' correction for small numbers. Patients and Methods

All operative records and discharge diagnoses for infants up to 1 year of age were reviewed for the years 1976 through 1982 at the Kosair Children's Hospital in Louisville, Kentucky. Patients entered into the study had gastrointestinal perforation identified at operation or autopsy, which included in utero perforations and abdominal wall defects when associated with perforations. Patients with peritonitis alone were excluded. There was a total of 29 infants (8 female and 21 male) with gestational ages of 26 to 43 weeks (Figure 1). Eight term infants (greater than 36 weeks gestation) weighed 2,100 to 4,300 g, (two were small for gestational age) and 21 premature infants weighed 675 to 2,500 g. Average Apgar scores 1 and 5 minutes after perforation were 5.2 and 7.3 in the term infants and 5.6 and 7.1 in the premature infants, and were not significantly different. All patients were less than 28 d a y s of age at the time of perforation except two. These patients, aged 35 and 46 days, resided in the neonatal intensive care unit from birth through their recovery from perforation. No older infants of up to 1 year of age were identified as having gastrointestinal perforation. This rarity of perforation occurrence during infancy is consistent with previous series [1-6]. Sixty-nine percent of perFrom the Departmentof Surgery, Divisionof Pediatric Surgery, University of LouisvilleSchoolof Medicineand KosairChildren'sHospital, Louisville, Kentucky. Requests for reprintsshouldbe addressedto AnthonyP. Borzotta, MD, Veterans AdministrationMedical Center, 10701 East Boulevard, Cleveland, Ohio 44106.

Volume 155, March 1988

forations occurred in the first week of life, 86 percent by 2 weeks, and 93 percent by 3 weeks. Twenty-two perforations occurred in the midgut. Multiple perforations in six infants occurred in long lengths of bowel involved with necrotizing enterocolitis or ischemic iatrogenic injuries. With a single exception, these multiple perforations clustered in one length of diseased bowel. Therefore, multiple perforations in a single patient were considered as one for statistical purposes. The perforations were classified by probable underlying cause in Figure 2. Of two in utero perforations in infants who were later found to have cystic fibrosis, one presented as acute meconium peritonitis with volvulus and another as a sterile mass resulting from a remote sealed perforation. Bowel ischemia in six infants was related to perinatal or later asphyxial events. Asphyxial events were defined as perinatal stresses, such as fetal heart rate decelerations, maternal hemorrhage, meconium aspiration, and immediate respiratory distress at birth or consequent to late diagnostic maneuvers which increased the risk of fetal hypoxia. The perforation on day 46 occurred 4 days after bronchoscopy and dilatation of a tracheal stenosis, involving three focal perforations in one segment of ileum. Necrotizing enterocolitis with midgut perforations developed in 10 premature infants and 1 term infant. Four infants had mechanical problems. Two had abdominal wall defects and two had distal bowel obstruction with proximal perforation. Puncture or iatrogenically induced bowel ischemia due to umbilical artery air or clot embolization occurred in four infants, two of whom died. Focal idiopathic perforations of the distal ileum and appendix developed in two premature infants at 4 and 35 days of life. Perforation of the distal ileum occurred against a background of respiratory failure. The biopsy specimen contained only hemorrhagic necrosis. Perforation of the appendix was associated with terminal ileal necrotizing enterocolitis, but the appendix was histologically uninvolved. Both infants survived. Four infants presented at birth with evidence of in utero or acute perforation. Later, symptoms of perforation included apnea and bradycardia (24 percent of the patients), diarrhea (17 percent), and bilious vomiting (14 percent). Signs reflecting developing peritonitis or pneumoperitoneum included distended abdomen (76 percent of the patients), acute or exacerbated respiratory distress

447

Borzotta and Groff

6-

4000. 3500 l

v LIVED 9 DIED

5

r-=.l I ] ALIVE

-

I

4 -

"~ 3000,

nrl.~ ~3-

E

DEAD

~zsoo z 2,

V

N

NEC

I

2

I

A = ASPHYXIAL EVENT I = IATROGENIC LU.= IN UTERO

I

M = MECHANICAL DEFECTS NEC = NECROTIZING ENTEROCOLITIS

~"-'J(/'-"~

I~//,J

V ~7 I

1500

u

'

v v 0

I000'

500 -

9 14 16

2'e 3'o 3'2 3'4 3'6 ~h TE'e. ~2 4'4 GESTATIONAL AGE (weeks)

Figure 1. Survival by birth weight and gestational age. Survival rates were significantly lower in Infants less than 1,000 g at blrth and 29 weeks of age or less.

(31 percent), thrombocytopenia (24 percent), and metabolic acidosis (17 percent). One infant presented with a palpable mass. Specific features of respiratory distress were labored respirations, cyanosis, tachypnea, and sudden decreases in arterial oxygen pressure. Worsening cardiac or pulmonary dysfunction led to diagnostic studies in 12 infants. Abdominal roentgenograms revealed pneumoperitoneum in 19 of 25 infants (76 percent). Pathogenesis:In term infants, there is a trend towards proximal sites of perforation (Figure 3, left) which coincides with a different spectrum of causation in comparison to premature infants (Figure 3, right). Four of eight term infants had perforations at or proximal to the ligament of Treitz. Each was associated with mechanical factors which probably increased intraluminal pressure. Two of the infants had gastroschisis, one infant had nongangrenous midgut volvulus, and one infant had spontaneous gastric rupture. Of the four distal perforations, one occurred in an infant with cystic fibrosis and ileal perforation in utero which had sealed and became an atretic segment, the second occurred in a gangrenous segment of a midgut volvulus, a third occurred in the only term infant with necrotizing enterocolitis, and the fourth resulted from catastrophic thrombosis of the entire abdominal aorta after umbilical artery catheterization. The infant with gastric rupture was in general good health up to the time of occurrence at 48 hours of age. Gastric rupture occurred at home shortly after discharge from the hospital, and the patient died from intractable shock after repair, as did the child with thrombosis of the aorta. T h i s 25 percent mortality rate in the term infants was lower than the 47.6 percent rate in the premature infants, but not significantly lower (chi-square = 1.22, p >0.1). It was related as much to delay in diagnosis and faulty management (possibly too early postnatal discharge and failure to remove an umbilical artery catheter when an infant's legs became ischemic) as to the disease processes themselves.

448

3

4

5

6

7

IO 15 DAY OF LIFE 8

9

,,

21

,,

35

,

46

Figure 2. Onset of symptoms and causes of perforation. Wfth few exceptions, diagnosis was made and treatment rendered within 24 hours of onset of symptoms. Three trends are noted: mechanical obstruction phenomena occurred at or soon after birth, both asphyxlal events and necrotizlng enterocolltls occurred frequently and sometimes occurred late, and latrogenlc Injuries clustered in the first week when Infants were most ill and monilorlng was most Intense. Question marks Indicate Idiopathic perforation. A : asphyxlal event; I = iatrogenic; IU = in utero; M = mechanical defects; NEC = necroUzing enterocolftLr

The influenceof bowel ischemia on the siteof perforation and outcome in the premature infants isapparent in Figure 3, right.Perinatal asphyxial events, Apgar scores of lessthan 6 at 5 minutes, low cardiac output states,and iatrogenicinjuriesalteringintestinalblood flow were associatedwith 81 percent of the perforations.The remainder were of idiopathic origin.Eighty-six percent of the perforations occurred in the ileocecalregion where the splanchnic circulation is most distant from its aortic source. Two small focal perforations in the stomach (10 percent) were of indeterminate cause but were associated with perinatal asphyxial events. They may have represented acute peptic ulcers complicated by perforation. Symptomatic patent ductus arteriosus syndrome may be associated with splanchnic low-flow states due to a reduction in the oxygen tension of arterial blood and by adverse alteration of the cardiodynamics and distribution of cardiac output. Ten premature infants had symptomatic patent ductus arteriosus prior to gastrointestinal perforation (Table I). One patient had spontaneous closure, nine were treated medically, and four eventually underwent ligation. No particular pattern of site or cause of perforation was associated with patent ductus arteriosus. The only surviving patient was treated by ligation of the duct. Three of five medically treated infants had a patent duct at autopsy. Eleven infants had necrotizing enterocolitis, and all but i were premature. Five of these infants died, for a mortality rate similar to that from nonnecrotizing enterocolitis related perforation in premature infants (chi-square = 0.043), indicating that the surgeon should be as vigorous treating necrotizing enterocolitis perforation as he is with perforation from other causes. All sites of perforation were in the ileocecal region. Perinatal complications, low 5 minute Apgar scores, and early respiratory problems were

The American Journal of Surgery

Gastrointestinal Perforat on in Infants

GASTRIC RUPTURE~I GASTROSCHISI$<3 VOLVULU$~

AORTIC THROHOOSI$

NEC?

+

IVOLVULU$

LIVED q DIEO __

Figure 3, Birth weights and causes and sites of perfora#on In term Infants (left) and premature Infants (right). Multiple perforations are Included. Question marks Indicate Idiopathic perforation. A = asphyxlal event; I = latrogenlc; NEC = necrotlzlng enierocolllls,

]LIVED IDLED . LENGTH OF INVOLVED 8GWEL

=

!

"!

i

i

2000 3600 4(300 BIRTH WEIGHT (groins)

no more closely associated with the development of necrotizing enterocolitis than with other causes of perforation in this population. Survival is significantly influenced by the impact of perinatal complications, degree of prematurity, persistent infection, and emergency reoperation. Twenty infants in the group had 40 maternal and perinatal complications, as described in Table II. Circumstances of a contaminated birth canal, such as amnionitis, cervical Mycoplasma ominus infection, premature rupture of membranes for 3 to 30 days (in one case exacerbated by precipitate delivery in an automobile), and maternal hemodynamic instability from vaginal bleeding, placenta praevia, and preeclampsia were associated with a 75 percent mortality rate in our patients (chi-square = 4.504;

TABLE I

LENGTH Of" INVOLVEOBOWEL

0

IO00

2000 5000 BIRTH WEIGHT (aroms)

0.02< p <0.05). We found that infants weighing less than 1,000 g had a significantly lower survival rate (20 percent) than did heavier newborns (78 percent) (chi-square = 7.128; 0.01> p >0.001), coinciding with a gestational age of greater or less than 29 weeks (Figure 1). Treatment: Infants had blood and spinal fluid specimens taken for bacterial culture and were started on a standard antibiotic regimen of intravenous ampicillin and gentamicin. Methicillin, kanamycin, or amikacin were added or substituted when necessary. Those pa: tients with a proved or suspected diagnosis of necrotizing enterocolitis were frequently receiving ampicillin and gentamicin prior to perforation. Laparotomy was indicated in 19 infants based on the finding of free air in the peritoneum on roentgenograms. Other indications for ab-

Patent Ductus Arterlosus In 29 Neonates With Gastrointestinal Perforations

Weight (g)

Medical Therapy

Day of Ligation

Patency at Autopsy

Outcome

900 700 1,940 1,000 790 900* 1,350 680 960 1,000

Indomethacin, Pfiscoline Indomethacin Priscoline Indomethacin Priscoline

:.. 11 ... ... ...

indomethacin Indomethacin Indomethacin Indomethacin

"1"0"

? Closed Open ? Open Closed Closed Open Closed Closed

Dead Dead Dead Dead Dead Dead Dead Dead Alive Dead

"1'21 3

* Closed spontaneously.

Volume 155, March 1988

449

Borzotta and Groff

TABLE II

Perinatal and Birth Complications in 29 Infants Complications

Maternal Premature rupture of membranes Birth canal infections Incompetent cervix Placenta praevia Vaginal bleeding Preeclampsia Unattended delivery (automobile) Fetal and Newborn Apgar score < 6 at 5 min Immediate respiratory distress Fetal distress in utero Gastroschisis Breech presentation, twin Sepsis (E. coil bacteremia) Meconium staining

TABLE IV

TABLE III

Term (n = 5)

Premature (n = 15)

0 0 0 0 0 1 0

3 2 2 1 1 0 1

2 2 2 2 0 0 1

9 7 1 0 1 1 1

Incidence of Abdominal Reoperation and Mortality After Gastrointestinal Perforation*

Operation

1

Elective Urgent Total

29 i:31) 29 (41)

Number of Operations 2 3 9 (0) 8 (33) 15 (14)

1 (0) 2 (0) 3 (0)

4 1 (0) 1 (100) 2 (50)

" Values in parentheses are mortality rates.

dominal exploration were deteriorating health despite maximal correction of acidosis, fluid and blood loss, and antibiotic coverage. The development of significant inflammation of the abdominal wall and continuing sepsis were also indications for surgical exploration. Operative findings and treatment are summarized in Table III. Four perforations of the stomach, one of the duodenum, and one of the jejunum were closed primarily. Tube gastrostomies were placed for all perforations proximal to the ligament of Treitz, except in one infant. Portions of ischemic bowel were resected in 1i infants and enterocutaneous stomas were created in 8 infants. Isolated distal focal perforations were closed primarily in three infants and exteriorized in three infants. Exploration without resection was performed in two infants with ischemic necrosis of the entire small bowel, which would have necessitated removal of the entire midgut and colon. One of these infants underwent a second-look laparotomy within 24 hours. Both patients died. Results

All 29 infants underwent at least one laparotomy. Five died within days from irreversible shock or uncorrectable lesions and five recovered without further operations. The survival rate for the entire

450

Primary Operation, Relaparotomy, and Mortality Mortality* Primary Reoperation

Operation Segmental intestinal resection End stoma Primary anastomosis Right hemicolectomy Closure of perforation Gastric Small bowel Exteriorization of perforation Appendectomy Exploration only

8 (38) 3 (33) 6 (33)

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

4 (75) 2 (0) 3 (33) 1 (0) 2 (100)

... 2 "-(0)' 1 (100)

* Values in parentheses are percentages.

group of newborns was 59 percent. The frequency of elective relaparotomy to reestablish continuity of the gastrointestinal tract and urgent operation for obstruction or infection is described in Table IV. Mortality rates for each procedure were calculated from outcomes up to the time of death, next reoperation, or hospital discharge. We believe that elective reoperation for closure of stomas is safe and urgent and that emergent relaparotomy, necessitated by continuing necrotizing enterocolitis, abscesses, or intestinal obstruction from adhesions, carries risks similar to the initial procedure. The data suggest that the risk in reoperation lies more with the disease process requiring operation than with the surgical procedure itself. Seventeen surgical complications developed in 13 infants. Continuing infection often precipitated death. However, the newborns who survived or escaped intraabdominal problems remained at risk for other complications. Sixteen initial survivors had 35 intercurrent or late nonsurgical complications (Table V). Typically, these were the consequences of prematurity or prolonged dependence on life-support systems and included intracranial hemorrhage and bronchopulmonary dysplasia. The more severe complications caused or significantly contributed to deaths in half of these patients. Long-term follow-up data are not available. Comments

Premature birth is characterized by incomplete development of pulmonary, cerebral, and hemodynamic systems, low birth weight resulting in decreased energy reserves, and increased susceptibility to a variety of complications. In the present Study, pneumonia developed in only one term infant who survived. All of the other complications occurred in premature infants. In eight of these infants, hemorrhage, infarction of the central nervous system, or chronic pulmonary disorders were the eventual cause of death. Thus, premature infants

The American Journal of Surge~

Gastrointestinal Perforation in Infants

TABLE V

TABLE Vl

Complications and Causes of Death

Complication

Infants Total Died

Cause of Death

Impact of Elective Operation and Atimentary Tract Perforation on Neonatal Survival (reported infant survival, 1976 to 1982)*

Surgica! and Postoperative Complications in 13 Infants Peritonitis & abscess Bowel Obstruction Wound infection Colocutaneous fistula Dehiscence Massive NaHCO3 loss Stomal stricture Ventral hernia

5 4 3 1 1 1 1 1

Weight (g)

3 2 1 1 1 0 0 0

500 io 1,001 to 1,501 to 2,001 to >2,500

1,000 1,500 2,000 2,500

Singleton Major Operated Live Births Elective Gastrointestinal (%) Operation (%)" Perforations (%)1" 36.7 8i.2 93.7 98.2 99.7

84 79 91 ... ...

22.2 67.1 75 100 66.7

9 Thoracotomy for patent ductus arteriosus and laparotomies

Nonsurgical Intercurrent Complications in 16 Infants

[lo, 11]. 91 This is a series of 29 cases.

Patent ductus artedosus Intracranial hemorrhage Seizures Pneumonia Bronchodysplasia & other Tracheal stenosis Congestive heart failure Hydrocephalus Ischemic colon stricture Sepsis, unknown site

10 5 4 4 4 2 2 2 1 1

9 4 2 2 4 2 1 1 0 1

2 2 3

1

NaHCO3 = sodium bicarbonate.

surviving the septic course Of perforation remain at high risk for late death due to disorders arising from prolonged ventilatory support and cerebral immaturity. The deferment of mortality into the postneonatal period has been described in a statewide review of infant mortality [6]. Since the 1939 review by Thelander [7l of newborn gastrointestinal perforations, several signifiCant changes have occurred [8,9]. First, willingness to operate has increased in conjunction with surgical technical advances and improvements in respiratory and metabolic support, blood transfusion capability, and the use of antibiotics: Second, the age and birth weight of infants with perforation have significantly decreased. But despite the increase in the number of premature infants, survival rates continue to improve. Fonkalsrud et al [2] reported 5 survivors among 50 premature infants weighing under 5 pounds (10 percent). Current survival figures are 37 percent for infants weighing a third of that limit. Elective operations at all birth weights have comparable safety, reflecting the effectiveness of modern support methods in stable infants (Table VI) [10,11]. Perforation in the presence of bacterial peritonitis, however, has the greatest deleterious effect in the very low birth weight group. Third, the pattern of causes Of perforation has Changed from (in :declining incidence) obstruction, spontaneous (peptic) trauma, and technical failure to necrotizing enterocolitis, perinatal asphyxial events, iatrogenic injury, obstruction, and Others [2,3,5]. The distribution of perforations has also shifted, moving distally Volume 155, March 1988

in the alimentary canal as new diseases afflict the group at greatest risk; namely, premature infants with hyaline membrane disease, patent ductus arteriosus, congestive heart failure, and bronchial dysplasia. Early series reported approximately equal numbers of foregut perforations compared with distal midgut and hindgut perforations [2,7,12113]. Currently, the ratio is 1:2 [5,14]. The two distributions by site probably reflect different, but not exclusive, mechanisms of injury. Indeed, multiple factors appear to contribute to the Occurrence and locatiOn of a given perforation. The present study indicates that although perforation of the intestine in an infant is a catastrophic event, the very premature infant is at significantly greater risk of dying. A substantial share of their mortality is related to intracerebral hemorrhage and bronchopulmonary dysp!asia which occur in this class of infants. Our survival rate for premature infants in the 1,000 to 2,000 g weight category was above 50 percent and demonstrates that, in a modern neonatal intensive care unit, the physicians and surgeons should use vigorous medical and surgical approaches when trying to successfully treat premature infants with gastrointestinal perforation: It should be possible to improve the survival rate of this age group in the future. Summary

We found that the mortality rate was no greater in patients with necrotizing enterocolitis complicated by perforation compared with the rate in those with other causes of perforation in similar weight classes. Iatrogenic injuries had a 50 percent mortality rate in all gestational age and birth weight categories and, therefore, great caution and surveillance are of utmost importance when using invasive monitoring and therapeutic modalities to prevent this cause of perforation. Term infants had a preponderance of mechanical causes of perforation which mainly occurred in the foregut and proximal midgut, whereas premature infants have a preponderance of asphyx: 451

Borzotta and Groff

ial or ischemic events underlying perforations which mainly occurred in the ileocolic region and were often associated with necrotizing enterocolitis. Although neonatal intestinal perforation is a catastrophic event, the very premature infant weighing less than !,000 g at birth is at significantly greatest risk. The discouraging 20 percent survival rate in the less than 1,000 g premature infants presents a challenge to the surgeon, since the overall Survival rate was 59 percent and the term infants had a 78 per~cent survival rate. A substantial share of the mortality in the infants weighing less than 1,000 g at birth relates to the occurrence of intracerebral hemorrhage and bronchopulmonary dysplasia. Vigorous medical and surgical approaches can be used to salvage premature infants in all Weight classes with gastrointestinal perforation. References 1. Cruze K, Synder WH. Acute perforation of the alimentary tract in infancy and childhood. Ann Surg 1961; 154: 93-9. 2. Fonklasrud EW, Ellis DG, Clatworthy HW. Neonatal peritonitis. J Pediatr Surg 1966; 1: 227-38.

452

3. Emanuel B, Zlotnik P, Raffensperger JG. Perforation of the gastrointestinal tract in infancy and childhood. Surg Gynecol Obstet 1978; 146: 926-8. 4. Wynne JM. Spontaneous neonatal and fetal Intestinal perforation. S Afr Med J 1977; 52: 351-3. 5. Bell MJ. Perforation of the gastrointestinal tract and peritonitis In the neonate~ Surg Gynecol Obstet 1985; 160: 20-6. 6. Buehler JW, Hogue CJR, Zaro SM~ Postponing or preventing deaths? Trends in infant surgical, Georgia, 1974 through 1981. JAMA 1985; 253: 3564-7. 7. Thelander HE. Perforation of the gastrointestinal tract of the newborn infant. Am J Dis Child 1939; 58: 371-93. 8. Birtch AG, Coran'AG, Gross RE. Neonatal peritonitis. Surgery 1967; 61:305-13: 9. McDougal WS, Izant RJ, Zollinger RM. Primary peritonitis in infancy and childhood. Ann Surg 1975; 181: 310-3. 10. Daum R, Schutze U, Hill E, Hoffman H. Mortality of preoperative peritonitis in newborn infants without intestinal obstruction. Prog Pediatr Surg 1979; 13: 267-71, 11. Seashore JH, Tou!oukian RJ, Kopf GS. Major surgery in infants weighing less than 1500 grams. Am J Surg 1983; 145: 483-: 7. 12. Freeark RJ, Raffensperger JG, Cond0n JB. Pneumoperitoneum in infancy. Surg Gynecol Obstet 1961; 113: 623-30~ 13. Lloyd JR. The etiology Of gastrointestinal perforations in the newborn. J Pediatr Surg 1969; 4: 77-84. 14. Shashikumar VL, Bassuk A, Pilling GP, Gresson SL. Spontaneous gastric rupture in the newborn. Ann Surg 1975; 182: 22-5.

The American Journal of Surgery