Audit of neonatal deaths of nonmalformed infants of 34 or more weeks' gestation: unavoidable catastrophic events or suboptimal care?

Acta Obstet Gynecol Scand 2002: 81: 899–904 Copyright C Acta Obstet Gynecol Scand 2002

Printed in Denmark . All rights reservedActa Obstetricia et

Gynecologica ScandinavicaISSN 0001-6349

ORIGINAL ARTICLE

Audit of neonatal deaths of nonmalformedinfants of 34or more weeks’ gestation:unavoidable catastrophic events or suboptimalcare?JAN HOLT1, INGEBJØRG FAGERLI1, BJØRN HOLDØ2 AND INGAR NIKOLAI VOLD3

From the 1Department of Pediatrics and 2Obstetrics and Gynecology, Nordland Central Hospital, Bodø, and 3Lofoten Hospital,Gravdal, Norway

Acta Obstet Gynecol Scand 2002; 81: 899–904. C Acta Obstet Gynecol Scand 2002

Background. The neonatal death rate (death 28days/1000 live births) has decreased andthe level is now so low that it has been questioned whether further improvement is possible.The aim of this study was to categorize nonmalformed infants of 34 weeks’ or moregestational age dying in the neonatal period to analyze if these deaths might have beenprevented.Material and methods. We used the audit method to study neonatal deaths during 1986–98in a county population of approximately 240000 inhabitants.Results. Twenty-six neonatal deaths from a population of 41901 live births were analyzed.The neonatal deaths were found to be associated with antepartum hypoxia (six cases);intrapartum catastrophes (seven cases); intrapartum monitoring deficiencies (five cases);resuscitation and stabilization after birth (two cases); infection (one case); sudden infantdeath syndrome (four cases); and peritonitis (died at home, one case). Suboptimal care wasrecorded in 16 cases. Neonatal death was unlikely to be associated with suboptimal care insix cases, but in 10 cases suboptimal care might or was likely to have brought about the fataloutcome.Conclusion. Avoiding suboptimal care might or is likely to prevent neonatal death in 10/26(38.5%) of nonmalformed infants of 34 weeks’ or more gestational age. Such improvementsmay, however, only slightly influence the neonatal death rate, with a reduction from 4.4 to4.2/1000 live births.

Keywords: audit, neonatal death, perinatal, suboptimal care

Submitted 9 August, 2001Accepted 10 January, 2002

Death rates have often been used to measure peri-natal outcome. The neonatal death rate (deathwithin 28days/1000 live births) in Norway has de-creased from 11.1/1000 live births in 1967 to 2.8/1000 live births in 1998. A mean value of 3.3/1000live births was calculated for the 1986–98 periodin Norway (1) vs. 4.4/1000 live births in Nordlandcounty in the same period.

It has been suggested that the number of ‘poss-ible preventable’ neonatal deaths according to di-agnosis is very small (2). Also, the rate for subopti-

C Acta Obstet Gynecol Scand 81 (2002)

mal neonatal care associated with death has beencalculated to be even low: 7% or 0.2/1000 livebirths (3). When taking into account the obstetric‘avoidable’ component, however, a figure for sub-optimal care of 1.1–1.8/1000 live births has beenreported (4,5).

Every live born infant dying in the neonatalperiod is included in these statistics. Preterm in-fants of 23–33weeks’ gestational age very oftensuffer from respiratory distress syndrome, intra-cranial hemorrhage, or other diseases character-

900 J. Holt et al.

istic for infants with low gestational age. Infantsabove 33weeks’ gestational age are much morelikely to survive the neonatal period. When exclud-ing infants dying from major malformations, themost probable causes of death among term- andnear-term live born infants may be hypoxic-ische-mic injury or infections, categories that might beprevented.

Non-malformed infants of 34 weeks’ gestationalage very infrequently die in the neonatal period.As the neonatal death rate seems to be higher inNordland county than for the rest of the country,we wanted to raise this question: Are neonataldeaths among term and near-term infants causedby silent events during the antenatal period andunpreventable catastrophic events, or is the careduring or after delivery suboptimal? To answer thisquestion we studied neonatal deaths of 34 weeks’or more gestational age from a county population.

Materials and methods

Subjects

Nordland county has a population of 240000 in-habitants. Although the population is scatteredover a great area, the socioeconomic standard isrelatively high, and virtually all inhabitants enjoythe benefit of a readily available health-care systemat public expense. Perinatal care is organized ac-cording to the principle of regionalization. A cen-tral hospital receives women with preterm labor orotherwise at risk from nine other delivery units inthe county, and a neonatal intensive care unit(NICU) provides tertiary care for preterm and sickinfants. A county perinatal committee has con-secutively characterized stillbirths and neonataldeaths from Nordland county (240000 inhabi-tants). To the data from 1986 to 1996 (6) wereadded cases from 1997 to 1998. Cases were re-ported to the committee from obstetric depart-ments and nursing homes in the county and fromthe NICU. This report considers live born infantsof 34 weeks’ gestational age dying in the neo-natal period. To make certain that every case hadbeen retrieved, the gathered data was comparedwith lists of neonatal infant deaths from the Medi-cal Birth Registry of Norway. Infants with majormalformations, metabolic anomalies, or chromo-somal disorders or syndromes causing neonataldeath were excluded.

Study protocol

Information about each case was collected fromhospital protocols, patient records at the delivery


units, the NICU, and other units outside thecounty if needed. Details concerning antenatalcare, delivery, neonatal care, and autopsy findings(if autopsy was performed) were recorded. In a fewcases information was obtained from general prac-titioners.

Routinely, a second-trimester (15–18weeks) ul-trasound examination was used to estimate thegestational age. In three cases the last menstrualperiod rule was used. Ultrasound-estimated fetalgrowth curves (7) were used to decide whether aninfant was large for gestational age (birth weight mean birth weight π 2SD) or small for ges-tational age (birth weight mean birth weight – 2SD).

Fetal heart rate was monitored with a cardiotoc-ograph (CTG). Two obstetricians (INV, BH),blinded to the outcome of the infants, reviewed in-dependently the CTG-tracings. The blinding of thereviewers was achieved by adding a number of con-trol CTG-tracings taken from singleton term in-fants with an Apgar score of 7 at 1min. Inter-observer agreement for this review was moderate(weighted kappa 0.53, 95% CI 0.18–0.87). The in-formation gained from the CTG analysis was usedin the audit discussions. Characterizations of theCTG-tracings were carried out according to re-search guidelines (8). Fetal distress was defined ac-cording to Huddlestone (9): ‘A precarious fetalcondition that, if allowed to persist, may lead topermanent damage or perinatal death’.

Clinical staging of hypoxic-ischaemic encephal-opathy (HIE) grade I–III was carried out accord-ing to Levene et al. (10) (Table I).

A perinatal audit process (11) was set up. To en-sure more objective analysis, all means of patientidentification and notes from midwives and doc-tors were removed from the presentations. Eachcase was discussed with intention to uncover ante-natal factors, catastrophic events, and suboptimalcare.

Intermediate outcome measures

O Cases most probably suffering from antenatalhypoxia expressed by observations and findingsbefore, at admittance to the delivery unit, orlater. Evidence of antenatal hypoxia was mainlybased on information concerning reduced fetalmovements or fetal heart rate bradycardia be-fore admittance, CTG-tracing at admittance(reduced FHR variability), and findings on cer-ebral imaging or autopsy dating an episode ofhypoxia back to the antenatal period.

O A major catastrophic event was defined as

Audit of nonmalformed neonatal deaths 901

Table I. Schematic clinical grading of hypoxic-ischaemic encephalopathy and suboptimal care factors

Grade Symptoms

Hypoxic-ischaemic encephalopathyI (mild) Irritability, hyperalert, mild hypotonia, poor suckingII (moderate) Lethargy, seizures, marked abnormalities of tone, requirement for tube feedingIII (severe) Coma, prolonged seizures, severe hypotonia, failure to maintain spontaneous respirationSuboptimal care factors0 No suboptimal factors have been identified1 Suboptimal factor(s) have been identified, but are unlikely to have caused the fatal outcome2 Suboptimal factor(s) have been identified, and might have caused the fatal outcome3 Suboptimal factor(s) have been identified, and are likely to have caused the fatal outcome

something that significantly worsened fetal oxy-genation for minutes or a few hours.

O Suboptimal care factors were identified if a sig-nificant deviation from adequate practice wasrecorded and/or an unnecessary delay of treat-ment occurred. A 0–3 grading of suboptimalcare was performed (12) (Table I).

Statistics

Inter-rater agreement was quantified by usingKappa statistics (MedCalc Software, 9030Mariakerke, Belgium). Comparisons betweengroups were made using the chi-square test. A p-value 0.05 was considered significant (InStatversion 3.0, GraphPad Software, Inc. San Diego,CA).

Results

During the study period there were 41901 livebirths among women residing in Nordland county.Sixty-two infants of 34 weeks’ gestational agedied during the neonatal period. Thirty-six ofthese infants were excluded for further analysis be-cause of major malformations, chromosomal dis-orders, or congenital metabolic defects: hypoplas-tic left heart syndrome (eight); trisomi 13 (five);other congenital heart defects (five); diaphragmatichernia (four); renal agenesis (three); cerebral mal-formations (two); trisomi 18 (two); multiple mal-formations (two); and others (five).

The neonatal death rate for infants of 34weeks’ gestational age was 1.48/1000 live births(0.86/1000 live births for malformed infants π0.62/1000 live births for nonmalformed livebirths).

Data on the basic characteristics of the 26 non-malformed infants and mothers are given in TableII. The number of cases varied between zero andfour per year. When comparing the number of newcases in 1986–92 (12 cases) to the number from


1993–98 (14 cases) there was no significant differ-ence between the two periods (pΩ0.6).

The 26 cases are characterized in groups accord-ing to the underlying starting event or time period.Suboptimal care factors and a grading of hypoxic-ischaemic encephalopathy and suboptimal care arealso given (Tables III–VI). Six cases were associ-ated with antenatal hypoxia, seven cases with in-trapartum catastrophes, and five cases with intra-partum monitoring deficiencies. Four infants suf-fered from sudden infant death syndrome. Two ofthese infants were born after caesarean section(previous caesarean section; maternal pelvis indi-cation), and in one case an autopsy was per-formed. Suboptimal care factors were recorded in

Table II. Basic characteristics of 26 mothers residing in Nordland county from1986–99 and 26 nonmalformed infants of 34 weeks’ or more gestational agedying in the neonatal period (Ω 28 days)

Characteristic n (SD)

MothersAge 35 years 2Smoking during pregnancy* 11Proteinuric hypertension 2Breech presentation 2Rupture of membranes 24 h 2Post-date pregnancy ( 42.0 weeks) 1Twin pregnancy 1

InfantsBirthweight (g) 3310 (815)Gestational age (weeks) 39.2 (2.2)Small for gestational age 3Large for gestational age 2

Apgar score at 5 min:0–3 124–6 77–10 7Convulsions during first 4 h 15Died day 1 6Died day 2–7 13Autopsy performed 14Died day 8–28 7Male: female 17 : 9

*Missing information in five cases.

902 J. Holt et al.

Table III. Neonatal deaths associated with antenatal hypoxia occurring before admission to the delivery unit: events and suboptimal care factors, and grading ofhypoxic-ischaemic encephalopathy and suboptimal care

Case Events and suboptimal care factors HIE Suboptimal care

1 FHR 96–104 recorded at home. Road ambulance III 1approximately 100 km. CTG: variable decelerations.Vacuum extraction. Meconium suctioned fromtrachea after 3–4 min MAS. Delay before removingmeconium from trachea.

2 Minimal FHR variability on admission CTG. MAS. III 1Delayed decision to perform a caesarean section.Pneumothorax.

3 Reduced fetal movements. CTG: minimal FHR variability. III 1Vaginal delivery.A caesarean section might have been performed 4 h earlier.

4 Ominous admission CTG. Vacuum extraction on a not III 1fully descended head. Shoulder dystocia. Infant diedafter 3.5 h. Meconium in membranes, cord and placenta.Hypertrophy of pulmonary arteries (autopsy) associatedwith meconium aspiration. Continuous scalp CTG tracing:caesarean section?

5 At antepartum control: irregular FHR. CTG: prolonged II 0decelerations at admittance. Day 3: cerebral atrophy (CT scan).

6 Oligohydramnios. Admission CTG: variable decelerations. II 0Day 4: White matter damage (CT scan).

FHR Ω fetal heart rate; CTG Ω cardiotocograph; MAS Ω meconium aspiration syndrome; HIE Ω hypoxic-ischaemic encephalopathy.

Table IV. Neonatal deaths with important contributions from intrapartum catastrophes: events and suboptimal care factors, and grading of hypoxic-ischaemicencephalopathy and suboptimal care


7 Suspected macrosomic foetus. Cord prolaps during examination. III 3Adverse treatment of cord prolaps (trial of vaginal delivery).

8 Breech presentation. Placental abruption III 2Growth retardation not recognized.Delayed decision to perform caesarean section.

9 Twin pregnancy. Preterm labor. Placental abruption. III 1Delay before caesarean section.

10 Breech presentation. Vaginal delivery. III 0Fracture of cervical vertebra and damage to the brain stem.

11 Antenatal bleeding, rupture of vasa praevia, severe fetal anaemia. III 012–13 Placental abruption. III 0

HIE Ω hypoxic-ischaemic encephalopathy.

16 cases (61.5%). In six cases the death was un-likely to be associated with suboptimal care. Sub-optimal care might have caused the fatal outcomeof five infants and was likely to have done so in afurther five cases.

Discussion

During the study period the neonatal death rateamong nonmalformed infants of 34 weeks’ ges-tational age in the present population was quitelow (0.62/1000 live births; 95% CI 0.41–0.91). Forthe 26 nonmalformed infants analyzed, 21 died inhospital and five at home. Among those dying inhospital, 20 infants suffered from hypoxic-ischemic


injury and all had early symptoms of moderate orsevere neonatal encephalopathy. Only one infantdied from a serious infection. As six cases mostprobably were associated with hypoxic events inthe antepartum period and seven cases were re-cruited from intrapartum catastrophes, these casesaltogether make up half of the whole group. So, ina birth population like ours with approximately3100 live births per year we might expect one ofthese events to occur each year. Perhaps, in the fu-ture, we should include most of these cases in anidealized neonatal death rate.

Suboptimal care factors were traced in 16 casesand consequential suboptimal care (grade 2–3) wasmost often associated with intrapartum monitor-

Audit of nonmalformed neonatal deaths 903

Table V. Neonatal deaths associated with intrapartum monitoring deficiencies: events and suboptimal care factors, and grading of hypoxic-ischaemic encephal-opathy and suboptimal care


14 Primiparous woman, post date pregnancy. III 3CTG: persistent severe variable decelerations.Dystocia (cephalopelvic disproportion). Vacuum extraction.Suboptimal communication midwife–obstetrician.A caesarean section might have been performed.

15 Primiparous woman. Dystocia (cephalopelvic disproportion). III 3Fetal distress. Delivered by forceps in a level I unit.Sub-aponeurotic and subdural hemorrhage.Referral to a level II delivery unit for caesarean section.

16 Moderate FHR variability, variable decelerations. III 2Velamentous insertion of the umbilical cord.Caesarean section. MAS.Growth retardation not recognized.Delayed contact midwife–obstetrician and delayed caesarean section.Pneumothorax.

17 Prolonged rupture of the membranes (36 h). III 2Fetal distress. Chorioamnionitis. Caesarean section.Caesarean section might have been performed 2 h earlier.

18 Primiparous woman. Dystocia (cephalopelvic disproportion). III 2Delayed referral to a level II delivery unit for caesarean section.

CTG Ω cardiotochograph; FHR Ω fetal heart rate; MAS Ω meconium aspiration syndrome; HIE Ω hypoxic-ischaemic encephalopathy.

Table VI. Infants experiencing adverse events mainly after birth and infants who died at home: events and suboptimal care factors, and grading of hypoxic-ischaemic encephalopathy and suboptimal care


19 Primiparous woman. Protracted delivery by vacuum extraction. III 3Infant died at 5 h. Delayed arrival of transport team.Severe subaponeurotic hemorrhage not observed.Improper documentation, monitoring and treatment.

20 Premature rupture of the membranes, uneventful delivery. III 3Apgar score 3 (1 min) and 3 (5 min). The infant did not breathe.Delayed and improper resuscitation.

21 Preeclampsia. Maternal fever. Preterm birth (34 weeks) 0 2Respiratory distress. Group B streptococcus septicaemia.Delayed start of antibiotic treatment.

22 Acute peritonitis (died at home). 0 023 SIDS (died at home). Growth retardation not recognized. 0 124–26 SIDS (died at home). 0 0

SIDS Ω sudden infant death syndrome; HIE Ω hypoxic-ischaemic encephalopathy.

ing deficiencies, handling of infants immediatelyafter birth, and with intrapartum catastrophes. Of21 neonatal deaths occurring in hospital, 10 casesmight or were likely to have been prevented. Itmight, thus, be feasible to consecutively analyzeneonatal deaths occurring at or near term with theintention to uncover personal or systemic malprac-tice.

Suboptimal care is not one single matter orevent. To show this we have, for each case, listedsome details including suboptimal care factors.For various reasons and in some cases it was re-lated to delayed treatment, which might be associ-ated with low self-efficacy among caregivers whodo not take action on observing fetal distress and


who do not follow up the effect of treatment im-plemented. In some instances, communication be-tween different professionals is not optimal. Ef-forts to reduce suboptimal care might thus beaimed at regularly using case stories for discussionof practical management among local caregivers.Hopefully, this might also initiate better communi-cation among team members.

Others (13,14) have reported nonattendance forantenatal care, noncompliance with medical ad-vice, and refusal to accept proposed interventionamong parturients. We did not record suboptimalcare factors of importance related to the motherherself. The way in which our data were collected,mainly from the delivery units and the NICU,

904 J. Holt et al.

made it possible to overlook suboptimal careamong parturient women during antenatal care.However, as the data was consecutively recordedand the cases were discussed in regular meetingstwice a year we estimate the probability of missingvaluable information to be low.

Smoking in pregnancy was recorded in 11 cases(Table II). As 30–40% of women smoke in preg-nancy, we have not characterized this informationas suboptimal care. Growth retardation was re-corded in three infants, none of these were knownbefore birth. We have characterized the handlingof these cases as suboptimal, although we areaware of the fact that few of these cases are tracedduring antenatal care (15).

Altogether, nine out of 21 women experiencingneonatal loss in hospital were at risk when arrivingat the hospital (Table II). This is contrary to an-other study reporting that neonatal encephal-opathy with fetal acidemia in term infants onlyfrom women at low risk and claiming that im-provements in obstetric care have focused very ef-fectively on high-risk populations (16).

Obstetricians might be biased by knowledge ofpoor neonatal outcome when retrospectively inter-preting fetal heart rate tracing and judging appro-priateness of obstetric care (17). We carried out theaudit analysis in two steps to try to minimize thistype of bias. We do not know whether the way inwhich the CTG-tracings were analyzed in the pres-ent study excluded bias, but this method was usefulin defining at what time appropriate interventionsmight have been given.

Information gained from the present study willbe given to caregivers at all delivery units and theNICU. As the population is rather homogenousand the standard of health care may be on a levelwith other areas, the findings in this study may befairly representative of other regions in Scandinav-ia. To explore the occurrence and effect of subopti-mal care factors further, prospective controlledstudies are needed.

Avoiding suboptimal care factors might or arelikely to prevent neonatal death in 10/26 (38.5%)nonmalformed infants of 34 weeks’ or more ges-tational age. Such improvements, although of ut-most importance in each specific case, might onlyto a small extent influence the neonatal death rate(reduction from 4.4 to 4.2/1000 live births).

References

1. Medical Birth Registry of Norway. Annual reports 1986–98. Medical Birth Registry of Norway, Bergen and NationalInstitute of Public Health, Oslo, Norway.


2. Philip AGS. Neonatal mortality rate: is further improve-ment possible? J Pediatr 1995; 126: 427–33.

3. Escobar GJ, Gardner MN, Chellino M, Fireman B, VerdiJ, Yanover M. Identification of neonatal deaths in a largecare organisation. Paediatr Perinat Epidemiol 1997; 11: 93–104.

4. Berge LN, Rasmussen S, Dahl LB. Evaluation of foetal andneonatal mortality at the University Hospital of Tromsø,Norway, from 1976 to 1989. Acta Obstet Gynecol Scand1976; 70: 275–82.

5. Moawad AH, Lee KS, Fisher DE, Ferguson R, PhillippeM. A model for the prospective analysis of perinatal deathsin a perinatal network. Am J Obstet Gynecol 1990; 162:15–22.

6. Holt J, Vold IN, Odland JØ, Førde OH. Perinatal deathsin a Norwegian county classified by the Nordic-Baltic peri-natal classification: geographical contrasts as a basis forquality assessment. Acta Obstet Gynecol Scand 2000; 79:107–12.

7. Marsal K, Persson P-H, Larsen T, Selbing A, Sultan B.Intrauterine growth curves based on ultrasonically esti-mated foetal weights. Acta Paediatr 1996; 85: 843–8.

8. National Institute of Child Health and Human Develop-ment Research Planning Workshop. Electronic foetal heartrate monitoring: research guidelines for interpretation. AmJ Obstet Gynecol 1997; 177: 1385–90.

9. Huddlestone JF. Intrapartum fetal assessment. A revue.Clinics in perinatology 1999; 26: 549–68.

10. Levene ML, Kornberg J, Williams THC. The incidence andseverity of post-asphyxial encephalopathy in full-term in-fants. Early Hum Dev 1985; 11: 21–6.

11. Dunn PM, McIlwaine G. Perinatal audit. A report pro-duced for the European Association of Perinatal Medicine.Prenat Neonat Med 1996; 1: 160–94.

12. Richardus JH, Graafsmans WC, Bergsjø P, Lloyd DJ, Bak-keteig LS, Bannan EM, Borkent-Polet M, Davidson LL,Defort P, Leitao AE, Langhoff-Roos J, Moralgarcia A, Pa-pantoniu NE, Wennergren M, Amelink-Verburg MP,Verldove-Vanhorlik SP, Mackenbach JP, the EuronatalWorking Group. Suboptimal Care and Perinatal Mortalityin 10 European Regions: Results of an international audit.Arch Dis Fetal Neonatal Ed 2002; 86: in press.

13. Biswas A, Chew S, Joseph R, Arulkumaran S, Anandaku-mar C, Ratnam SS. Toward improved perinatal care – peri-natal audit. Ann Acad Med (Singapore) 1995, submitted;24: 211–7.

14. Pattinson RC, Makin JD, Shaw A, Delport SD. The valueof incorporating avoidable factors into perinatal audits.SAMJ 1995; 85: 145–7.

15. Backe B, Nakling J. Effectiveness of antenatal care: a popu-lation based study. Br J Obstet Gynaecol 1993; 100: 727–32.

16. Westgate JA, Gunn AJ, Gunn TR. Antecedents of neonatalencephalopathy with fetal acidaemia at term. Br J ObstetGynaecol 1999; 106: 774–82.

17. Zain HA, Wright JW, Parrish GE, Diehl SJ. Interpretingthe fetal heart rate tracing. Effect of knowledge of neonataloutcome. J Reprod Med 1998; 43: 367–70.

Address for correspondence:Jan HoltDepartment of PediatricsNordland Central HospitalN-8092 BodøNorwaye-mail: Jan.Holt/nss.nl.no

Documents

Audit of neonatal deaths of nonmalformed infants of 34 or more weeks' gestation: unavoidable catastrophic events or suboptimal care?