PROSPECTIVE EVALUATION OFCLINICAL SCORING SYSTEMS IN INFANTS WITH BRONCHIOLITIS ADMITTED TO THE INTENSIVE CARE UNIT

Rödl S1, Resch B2, Hofer N2, Marschitz I3, Madler G1, Eber E4, Zobel G1

1 Paediatric Intensive Care Unit, Department of Paediatrics, Medical University of Graz, Austria

2 Research Unit for Neonatal Infectious Diseases and Epidemiology, Medical University of Graz, Austria

3 Department of Neonatology, ParacelsusMedicalUniversity of Salzburg, Austria

4 Respiratory and Allergic Disease Division, Department of Paediatrics, Medical University of Graz, Austria

Corresponding author’s address:

Univ. Professor Dr. Bernhard Resch, MD

Research Unit for Neonatal Infectious Diseases and Epidemiology

Division of Neonatology, Department of Paediatrics

MedicalUniversity of Graz

Auenbruggerplatz 34/2, 8036 Graz, Austria

Phone: 0043 316 38581134

Fax: 0043 316 38512678

Email:

Running title: Clinical scoring systems in bronchiolitis

Funding: None

Competing interests: None

The study was approved by the ethic committee of the Medical University of Graz

Abstract

Objective: To compare different scoring systems to assess severity of illness in infants with bronchiolitis being admitted to atertiarypaediatric intensive care unit (PICU).

Patients and Methods:Over an 18-year time period (1990 – 2007) infants with bronchiolitis aged up to 12 months and being admitted to the PICUwere prospectively scored using thePaediatric Risk of Mortality (PRISM III) score, the Organ System Failure (OSF) score, and the Acute Physiologic Score for Children (APSC)within 24 hours.Infants were compared regarding whether or not bronchiolitis was associated with RSV.

Results: There was no difference between 113 RSV positive and 80 negative infants regarding gestational age, birth weight, rate of premature delivery or BPD. The PRISM III score differed significantly between RSV positive and negative cases (3.27±0.39 vs. 1.96±0.44, p=0.006) as did the OSF score (0.56±0.05 vs. 0.35±0.06, p=0.049) and the APSC (5.16±0.46 vs. 4.1±0.53, p=0.048). All scores were significantly higher in the subgroup with mechanical ventilation (p<0.0001). The mean time of ventilation was significantly higher in the RSV positive compared to the RSV negative group (6.39±1.74 days vs. 2.4±0.47 days, p<0.001).

Conclusions: Infants suffering from RSV positive bronchiolitis had higher clinical scorescorresponding with the severity of bronchiolitis.

Key words: Bronchiolitis;clinical scoring systems; respiratory syncytial virus; intensive care unit
Introduction

Since the early 1980s various scoring systems have been used in paediatric intensive care units (PICU) to evaluate severity of illness. The Paediatric Risk of Mortality (PRISM) score was developed from the Physiologic Stability Index (PSI) by reducing the number of measured variables. This score was introduced by Pollack et al. in 1988 and adapted to PRISM III in 19961-3. The Organ System Failure (OSF) score exactly describes the criteria for failure of specific organ systems. It was introduced by Wilkinson et al. in 19874.The Acute Physiologic Score for Children (APSC) was derived by adapting variables ofAcute Physiology and Chronic Health Evaluation II (APACHE II) and the Simplified Acute Physiologic Score (SAPS) systems for infants and children5-7.The PRISM score was shown to require rigorous specific training and strict adherence to guidelines for reliable assessment8,9.Despite the excellent predictive power of these scores in PICU patients, no data concerning their predictive power of severity of illness in specific diagnoses such as acute viral bronchiolitis exist.

Common definitions of acute bronchiolitis comprisea viral respiratory tract infection characterizedby coryza and evidence of lower respiratory tract involvement manifested as tachypnoea, subcostal recession, and hyperinflation. North Americanclinicians would suggest that wheeze is the central feature whereas in the UK the central feature is suggested to be crepitations throughout the lung fields, with or without wheezing10.Theneed for supplemental oxygen and mechanical ventilation reflects the most severe forms of acute viral bronchiolitis.

A respiratory scoring system was developed for study purposes to assess severity of disease that was based on changes in oxygen saturation and respiratory rate and on pulmonary findings of retractions, wheezing, and crackles11. This scoring system was modified and simplified to the lower respiratory illness/infection score and characterized well more severe disease forms associated with respiratory syncytial virus (RSV) infection12,13. Although a specific score for RSV infection has been developed11-13, it is not feasible to use different scores for different diagnoses. In contrast,a severity of illness score is more appropriate for a general documentation of all PICU patients.

To our knowledge, no studies have been carried out to evaluate the scoring systems PRISM III, OSF, and APSC in severe acute viral bronchiolitis requiring admission to the PICU,focussing on the influence of RSV.The aim of this study was to evaluatethe severity of illnesscharacteriszed as need for ventilatory support in infants with RSV positive and RSV negative bronchiolitis using different clinical scoring systems and to compare these scoring systems.

Patients and methods

This study was part of a prospective surveillance of different clinical severity scores used in paediatric patients hospitalized at the PICU of a tertiary care paediatric centre between January 1990 and December 2007.The PRISM III, OSF, and APSC scoring systemswerecalculatedfor every paediatric patient admitted to the PICU during the first day of stay as described elsewhere1-6. Scoring was done by two consultants (S.R. and G.Z.) throughout the study period and documented using a software template incorporated in a proprietary electronic patient data management base. By change of the software from an ASCII file system and Turbo Pascal front end system to a Microsoft SQL Server database with a Delphi front end the PRISM Score was rescored to the PRISM III from the original values of the database.

All infants below 12 months of age withacute viral bronchiolitis according to the common European definitions14,and admitted to the PICU of the Department of Paediatrics of the Medical University of Graz, Austria, were included. Patients with underlying chronic diseases such as chronic pulmonary disease, airway anomalies, congenital heart disease, neuromuscular diseases, immunodeficiency, and chromosomal aberrations were excluded from this study group. Patients with recurrent wheezing episodes were also excluded.Criteria for admission to the PICU included clinical deterioration,i.e.increasing respiratory distress and/or increasing pCO2 values and/or evidence of apnoeas or cyanosis despite supplemental oxygen.Clinical data of the patients were recorded including age at admission, history of premature birth, gestational age, birth weight, history of bronchopulmonary dysplasia (BPD), duration of PICU stay, need for supplemental oxygen, ventilatory support with defined as continuous positive airway pressure or conventional mechanical ventilation, and secondary complications.All infants were tested for RSV and diagnosis of RSV was confirmed by RSV enzyme-linked immunosorbent assay antigen testing (Directigen™ RSV Test, Becton Dickinson, NJ, USA; sensitivity 97%, specificity 97%) from nasopharyngeal aspirates or wash. To analyse the influence of age infants were divided into four groups. Group 1: 0 – 30 days, group 2: 31 – 90 days, group 3: 91 – 180 days, and group 4: more than 180 days.

Our therapeutic strategies were monitoring, iv fluid, supplemental oxygen, ventilatory support with continuous positive airway pressure or conventional mechanical ventilation, and high frequency oscillation,heliox or extracorporeal membrane oxygenation in case of pulmonary deterioration. The mainly symptomatic therapy did not change over the study period.

Statistical analysis was performed usingthe Mann-Whitney-U-Test as appropriate for categorical data and the t-test as appropriate for numerical data. A p-value < 0.05 was considered to be significant. Analysis was done using Microsoft EXCEL (MS Office 2007) and StatView 4.5 (Abacus Concepts' statistical software 1999). Data are given as mean ± standard deviation. Receiver operator curves (ROC) were calculated for each score with regard to the need for ventilatory support. Data are presented as Area under the curve with 95% CI (confidence intervals) and specificity and sensitivity were calculated for optimal cut-off values.

Informed consent for statistical data analyses was providedfor every patient and the studywas approved by the local ethics committee.

Results

During the study period 1305 infants with acute viral bronchiolitis were hospitalized at the Paediatric Department.Out of these patients,193 infants younger than 12 months were admitted to the PICU due to severe illness according to our inclusion criteria. One hundred and thirteen infants (58.5%) were tested RSV positive and 80 (41.5%) negative. Twenty-four infants (12.4%) had a history of premature delivery and three infants (1.6%) had BPD. There was no difference between RSV positive and negative infants regarding gestational age and birth weight (mean 32.7 vs. 33 weeks, p=0.391, and 1882 vs. 2049 grams, p=0.246, respectively).Rate of premature delivery in RSV positive and RSV negative groups was 21% and 13%, respectively. BPD was found in 2.5% of the RSV positive infants and in 0.8% of the RSV negative ones. All our patients survived.

To evaluate the trending over the years we compared the scores of the first half of the study period with the second half of the study period as depictured in fig 1. No significant differences were observed.

Distribution of infants according to age groups 1-4was 35 (18%), 105 (55%), 41 (21%) and 12 (6%), respectively,demonstratedthe majority of infants being younger than 90 days.There were no significant differences in clinical scores between different age groups.APSC points in group 1-4 were 3.4±3.3, 4.9±4.8, 4.4.±4.77, and 6.2±5.0, respectively. OSF points were 0.4±0.5, 0.5±0.6, 0.4±0.6, and 0.2±0.4. PRISM III points were 2.8±3.8, 2.7±4.0, 2.2±3.9, and 2.0±3.0.Thus, the scores showed a good age adaptation.In addition, there was no significant difference in duration of ventilatory supportin groups 1-4.Days on ventilation calculated from entire groups were 2.71±3.82, 3.97±5.31, 3.63±7.15, and 1.5±3.38 in the fourage groups. The percentages of patientswith ventilatory support were 46%, 49%, 39%, and 20% in the fourage groups. Length of PICU stay did not change over time by calculating data for the first and second half of the study period (data not shown).

Eighty-seven infants (45%)needed ventilatorysupport, 47 infants (54% of the ventilated group)with a duration between 1 and 7 days, 30 (34.5%) between 8 and 14 days, and 10 (11.5%)for more than 14 days (median 7 days for the whole group). None of these infants needed heliox or extracorporeal membrane oxygenation. The ROC concerning need for ventilatory support showed a high significance for all three scores (p<0.001), the area under the curve was 0.87 for APSC (95% CI 0.81 – 0.91), 0.90 for OSF (95% CI 0.86 – 0.94), and 0.80 (95% CI 0.73 – 0.85) for PRISM III.The cut off point of 4 APSC points showed a sensitivity of 0.84 and a specificity of 0.77. The cut off point of 1 OSF point showed a sensitivity of 0.88 and a specificity of 0.94. The cut off point of 2 PRISM III points showed a sensitivity of 0.66 and a specificity of 0.87.

The percentage of patients with ventilatory support was significantly higher in the RSV positive group (51%) than in the RSV negative group (36%) (p<0.001). Furthermore,the mean time of ventilatory support differed significantly in both groups (6.39±1.74 days and 2.4±0.47 days, p<0.001). According to the age distribution of RSV infections we observed no significant difference; the percentages of RSV positive infantswere 57%, 64%, 46%, and 50% in the 4 age groups, respectively.

The PRISM III score differed significantly between RSV positive and negative infants(3.27±0.39 vs. 1.96±0.44, p=0.006) as did the OSF score (0.56±0.05 vs. 0.35±0.06, p=0.049) andthe APSC (5.16±0.46 vs. 4.1±0.53, p=0.048) (see fig 2).Comparing the ventilated RSV positive and the ventilated RSV negative infants scores did not revealany significant difference as shown in fig 3.

Bacterial co-infection was observed in 28% of RSV positive and in 31% of RSV negative infants. We found no significant difference concerning the rate of co-infections between these groups

All scores were significantly higher in the group ofmechanically ventilated patients (p<0.0001) (see fig 4).

Discussion

In this prospective study on the assessment of the three scoring systems, PRISM III, OSF and APSC in infants younger than 12 months admitted to a PICU due to acute viral bronchiolitis,significantly higher scores were calculatedin infants tested RSV positive. Thus, RSV led to more severe illness even in the setting of the PICU including only cases with severely deteriorating lower respiratory tract illness. Results were not influenced by age, gestational age, birth weight, BPD or bacterial co-infections154. As far as this study was part of a prospective surveillance, the problem with missing data or misclassification might be neglectable.

Clinical scoring systems provedto be useful for assessing severity of illness, care requirements, efficacy of therapy, and prognosis. The ideal score has to be simple, all variables must be measurable in every ICU, and it should be valid for a wide range of diagnoses and various age groups. Despite specific scores for specific diseases (e.g. meningococcal disease, lower respiratory tract illness) these physiological scoring systems accurately describe severity of disease in critically ill paediatric patients5.

The advantage of a general clinical score is the availability for all PICU patients.On the other hand, specific scores show a better description of one specific diagnosis but cannot be used for patients with other diagnoses. Inter-observer reliability of scoring systems has been questioned, but the data we report were acquiredby onlytwo experienced paediatric intensivists (S.R., G.Z.), whocalculated the scores during daily practice throughout the study period8,9.

PRISM IIIis the most common clinical score applied to PICU patientswith different diagnoses2.It contains 14 variables, three of them describing the respiratory system. It is very easy to handle and has also been used as a prognostic score for PICU patients.APSCcontains 17 variables including three of the respiratory system and has only been described as a clinical score for infants and children by our group5.OSFdefines the failing of seven organ systems including the respiratory system4.

With the change of the documentation software it was possible to recalculate the PRISM to PRISM III,but as other paediatric scores such as PELOD or PIM 2 were onlyintroduced later it was not possible to change the database and software to include them, too165,176.

In our study, PRISM III scoresmore clearly differentiated betweenRSV positive and RSV negative infants and between infants with and without mechanical respiratory support than the APSC or the OSF score. This result corresponds to the findings ofMidulla et al.18 who showed a more severe course of illness in infants with RSV infection with or without additional human bocavirus infection17.Furtheron RSV infection was proven to influence the neuroendocrine system and lymphocytes198,2019. A more severe course of disease in RSV positive infants was demonstrated not only by rates of ventilation but also by duration of ventilation, duration of admissions and last but not least the points of the three scoring systems.

As children show age dependent changes of physiological parameters, specific paediatric scores should account for these changes. PRISM III and APSC scores did not show any differencesforthe four age groups. In contrast to data in the literature, we could not find that younger infants had a more severe course of their respiratory illness than older infants210,221. Although we did not observe significant differences regarding duration of ventilation in all four age groups there was a tendency to longer durations of ventilation in infants younger than 3 months compared to those being older than 6 months. This might be explained by the referring strategy to our PICU of infants with acute viral bronchiolitis. Younger infants will be more rapidly referred to the PICU for cardiorespiratory monitoring than older infants232.To exclude the influence of pre-existing morbidities, patients were excluded from the study according to relevant exlusion criteria; other important factors proved to be evenly distributed in all groups243.

PRISM III is the most common paediatric score, and showed thebest discrimination. APSC has more parameters and showed less discrimination between the RSV positive and RSV negative group and between the groups with and without ventilatory support.The OSF score showed less significance andhadsless accurate description.

In conclusion, this is so far the first study demonstrating significantly higher PRISM III score, OSF score and APSC in infants below 12 months with acute viral bronchiolitis tested RSV positive compared to those tested RSV negative, hospitaliszed at a PICU.Infants with mechanical ventilation had significantly higher scores than those without.Respiratory and cardiac failure at admission led to high scoring values, which accurately corresponded with severity of bronchiolitis characteriszed as need for ventilation.The advantages of the PRISM III score are its significance, fewer variables included than the other scores and the ease of use. The PRISM III score appeared to be the ideal score to document severity of illness in infants with bronchiolitis.

References

1 Pollack MM, Ruttimann UE, Getson PR (1988)Pediatric risk of mortality (PRISM) score. Crit Care Med 16:1110-1116.

2 Pollack MM, Patel KM, Ruttimann UE (1996)PRISM III: an updated Pediatric Risk of Mortality score. Crit Care Med 24:743-752.

3 Pollack MM, Patel KM, Ruttimann UE (1997) The Pediatric Risk of Mortality III--Acute Physiology Score (PRISM III-APS): a method of assessing physiologic instability for pediatric intensive care unit patients. J Pediatr 131:575-581.

4 Wilkinson JD, Pollack MM, Glass NL, Kanter RK, Katz RW, Steinhart CM (1987)Mortality associated with multiple organ system failure and sepsis in pediatric intensive care unit. J Pediatr 111:324-328.

5 Zobel G, Kuttnig M, Grubbauer HM, Rödl S (1990)Evaluation of clinical scoring systems in critically ill infants and children. Clin Intensive Care 1:202-206

6 Zobel G, Kuttnig M, Ring E, Grubbauer HM (1990) Clinical scoring systems in children with continuous extracorporeal renal support. Child Nephrol Urol 10:14-17

7 Zobel G, Rödl S, Rigler B, Metzler H, Dacar D, Grubbauer HM, et al (1993) Prospective evaluation of clinical scoring systems in infants and children with cardiopulmonary insufficiency after cardiac surgery. J Cardiovasc Surg (Torino) 34:333-337

8 van Keulen JG, Polderman KH, Gemke RJBJ (2005)Reliability of PRISM and PIM scores in paediatric intensive care. Arch Dis Child 90:211-214

9 van Keulen JG, Gemke RJBJ, Polderman KH (2005)Effect of training and strict guidelines on the reliability of risk adjustment systems in paediatric intensive care. Intensive Care Med 31:1229-1234