Internationally lower respiratory infection (LRI) is a major cause of hospital admission in young children. Admissions for LRI have been increasing over the late 1990s and early 2000s in developed countries.1-3New Zealand has high rates of admissions for LRI and these admissions are concentrated in Māori and Pacific and in areas of high deprivation. Counties Manukau District Health Board (DHB) has the highest rate of admission for childhood LRI of any DHB in New Zealand.4LRI has been identified as a more useful epidemiologic and clinical description than differentiating between bronchiolitis and pneumonia, as there is significant variation in diagnosis and interpretation of chest X-ray with considerable proportions of children with a discharge diagnosis of bronchiolitis being found to have radiological evidence of pneumonia.3This study was undertaken to provide detailed prospective epidemiology in order to understand these admissions better. It formed part of baseline data collection prior to the introduction of conjugate pneumococcal vaccination (PCV7) to allow evaluation of its effect on LRI admission rates amongst a group at highest risk of pneumococcal disease.Methods Study population—Counties Manukau District Health Board (CMDHB) is responsible for approximately 433,000 people of whom 112,500 are 0-14 years of age including 30% European, 26% Pacific, 23% Māori, 15% Asian and 5% other or not stated.5 The CMDHB Middlemore Hospital includes Kidz First Children's Hospital which provides secondary acute emergency and inpatient care for children aged 0-14 years from the CMDHB geographic area. Ninety-five percent of children resident in the CMDHB area who are admitted with LRI are admitted to Kidz First. Children aged less than 2 years admitted to Kidz First with a primary diagnosis of LRI from 1 August 2007 to 23 December 2007 were eligible. Those admitted with a clinical diagnosis of a lower respiratory infection (bronchiolitis, pneumonia, bronchopneumonia, bronchitis, empyema/lung abscess/TB/pertussis/pneumonitis ) (ICD 10 bronchiolitis, ICD 9466.1, ICD 1010 J 21, pneumonia ICD 948486, 487, ICD 10 J 12-18, J 100 and J110, acute and specified lower respiratory infection ICD 10 J 22, acute bronchitis ICD 9 466, ICD 10 J20, lung abscess and parathorax ICD 9 510, 513, ICD 10 J 85-86, pertussis ICD 9 033, ICD 10 A37) were included. Study procedures—Eligible patients were identified as part of daily screening of admissions by the study nurse and consent was obtained from parents for study participation. Detailed information on household characteristicswas sought in a face to face parent questionnaire which included information on family, maternal age, housing, heating, smoke exposure and number in the house. Number of rooms in the house excluded bathrooms, showers, toilets, laundries, halls, garages and pantries. Socioeconomic deprivation for each child was estimated using the NZDep20066index for area of their residential address. The NZDep2006 index combines 9 variables from the 2006 NZ Census. Individual area scores are then ranked and placed on an ordinal scale from 1 to 10, with decile 10 representing the most deprived 10% of small areas. Children were clinically reviewed daily by the study nurse and the severity of illness documented. Nasopharyngeal aspirates were performed for viral and bacterial testing (data not presented in this paper.7) Relative risk of subgroups for at least one admission was calculated using the CMDHB 2007 birth cohort (data from National Minimum Dataset) as the denominator. In those admitted to hospital with LRI an analysis of risk factors related to total length of hospital stay in the 5 months of the study was performed. A general linear model was fitted with the log of the total length of stay (the sum of all admissions) for a child as the outcome and child's age (at start of study), mothers age, ethnicity of child, deprivation index of home address (coded as 1-8 or 9,10) whether or not premature (<36 weeks), household density (number of rooms in house divided by the number of people in the house), number of smokers in the home and whether or not there was a source of heating in the house included as explanatory variables. Analyses were also run with these variables included individually to ensure associations were not being obscured because of correlation among these explanatory variables. The study was approved by the Northern Regional Ethics Committee (NTX/07/07/059). Results During the study period there were 465 children with 580 admissions to Kidz First wards with LRI. Of these, 394 children (85%) with 508 admissions consented to the study. These admissions were similar to the total group in ethnicity and diagnosis. Table 1. Characteristics of study children admitted to CMDHB facilities with LRI, 1 August 2007-23 December 2007 Variables Cohort N=394 % CMDHB births 2007 n=8833 % Ethnicity Māori Pacific Other 118 221 55 29.9 56.1 14.0 1962 2808 4063 22.2 31.8 46.0 Deprivation 1 (least) to 8 9 10 (most deprivation) 84 124 186 21.3 31.4 47.2 3990 1747 3066 45.2 19.8 34.7 Prematurity Less than 33/40 33/40 to 35+6 36/40 or more 25 33 336 6.3 8.4 85.3 107 276 8309 1.2 3.1 94.1 Table 1 compares the characteristics of the study population with that of the 2007 birth cohort. Relative risk for admission was 4.4 (95%CI 3.2-6.2) for Māori, 5.8 (95%CI 4.4-7.9) for Pacific peoples compared with others, and 3.1 (2.4-3.9) for those living in the most deprived quintile (NZ Dep 9 and 10) compared with those in NZ Dep 1-8 areas. Relative risk of admission was 5.8 (3.8-8.7) for those born at less than 33 weeks gestation and 3.0 (2.1-4.3) for those born at 33<36 weeks gestation compared with those born after 36 weeks. Of the 508 admissions, 323 (64%) had a clinical diagnosis of bronchiolitis, 132 (26%) pneumonia, and 53 (10%) other LRI. Seventy-one (18%) had more than one admission and number of admissions ranged from 1 to 6. Length of stay ranged from 1 to 27 days with a median of 3 days. There was a total of 121 ICU bed days in 21 (5%) patients. The characteristics of households are detailed in Table 2. It is of note that two-thirds of children were exposed to smoke in their home and that two-thirds of families had English as their first language with most others speaking Pacific languages (Table 2). The median number of people in the house was 6, with range 2-31. Median number of adults was 2, median number of children 0-14y was 3, and 100 children (25%) lived in households with ≥7 other people. Thirty-three percent lived in households with four or more children. 224 (57%) had some period of full breastfeeding, 86 (22%) were partially breastfed. Twenty-seven percent lived in houses where caregivers reported never using any form of heating (Table 3). Thirty-nine percent of Pacific families compared with 11% of Māori and 11% of other families had no source of heating. Table 2. Characteristics of households of study children admitted with LRI to CMDHB, 1 August 2007-23 December 2007 Maternal age N % <20 yrs 20-24 yrs >25 yrs 60 117 217 15.2 29.7 55.1 Birth order 1st child 4th or subsequent 125 98 31.7 24.9 First language English Pacific Asian Māori Child in house in daycare No one in house in employment 266 113 8 5 111 44 67.5 28.7 2 1.2 28 11 Access to car during day Yes No 336 58 85.3 14.7 Access to phone Yes No 380 14 96.4 3.6 Smoking exposure Yes No 257 137 65.2 34.8 Table 3. Heating source for households with study children aged <2 years admitted with LRI to Kidz First Children's Hospital, 1 August 2007-23 December 2007 Heating source N % None Electric Bottled gas Wood Mains gas Coal Solar Other 106 210 68 54 10 5 1 3 26.9 53.3 17.3 13.7 2.5 1.3 0.3 0.8 Therapy given during the admission is illustrated in Table 4. Table 4. Therapies received by study children <2 years admitted with LRI to Kidz First Children's Hospital 1 August 2007 to 23 December 2007 by admission Therapies N % Respiratory Oxygen CPAP Ventilation 361 24 2 71.0 4.7 0.0 Fluids NG fluids IV fluids 234 118 46.1 23.2 Medications Ventolin Adrenaline Atrovent 140 34 14 23.2 6.7 2.7 Reflecting severity, 5% of admissions received CPAP and 87% (n=441) received oxygen, salbutamol or supplementary fluids. In 57% (288) of the admissions supplementary fluid was given either via NG tube (46%) and/or as intravenous fluids (23%). In 224 admissions a full blood count was performed, 56 (25%) had an abnormal haemoglobin. A primary care doctor was visited prior to 397 (78%) admissions. Following 390 (77%) admissions some form of follow-up was recommended after discharge, with 271 (53%) referred to the short-term Kidz First home nurse visiting service. In those admitted to hospital the total days in hospital during the study period was associated with prematurity (p=0.005), ethnicity (p=0.02) and age of child (p=0.0002) with those who were premature, or Māori or Pacific ethnicity likely to have a stay approximately 40% longer than respectively those not premature or of European ethnicity. The model also predicted every increase of a month in the child's age to result in a 3% reduction in length of stay (see Table 5). The estimates of the effect sizes differed only marginally when the explanatory variables were analysed individually, with no change in the variables able to be shown to be associated with length of stay. Table 5: Regression coefficient estimates for the log of total length of stay (the sum of all admissions) for a child as the outcome and child's age in days (at start of study), mothers age in years, ethnicity of child, deprivation index of home address (coded as 1-8 or 9,10) whether or not premature (<36 weeks), household density (number of rooms in house divided by the number of people in the house), number of smokers in the home and whether or not there was a source of heating in the house included as explanatory variables.. Source Estimate SE* p value Ethnicity Māori v Pacific

Internationally lower respiratory infection (LRI) is a major cause of hospital admission in young children. Admissions for LRI have been increasing over the late 1990s and early 2000s in developed countries.1-3New Zealand has high rates of admissions for LRI and these admissions are concentrated in Māori and Pacific and in areas of high deprivation. Counties Manukau District Health Board (DHB) has the highest rate of admission for childhood LRI of any DHB in New Zealand.4LRI has been identified as a more useful epidemiologic and clinical description than differentiating between bronchiolitis and pneumonia, as there is significant variation in diagnosis and interpretation of chest X-ray with considerable proportions of children with a discharge diagnosis of bronchiolitis being found to have radiological evidence of pneumonia.3This study was undertaken to provide detailed prospective epidemiology in order to understand these admissions better. It formed part of baseline data collection prior to the introduction of conjugate pneumococcal vaccination (PCV7) to allow evaluation of its effect on LRI admission rates amongst a group at highest risk of pneumococcal disease.Methods Study population—Counties Manukau District Health Board (CMDHB) is responsible for approximately 433,000 people of whom 112,500 are 0-14 years of age including 30% European, 26% Pacific, 23% Māori, 15% Asian and 5% other or not stated.5 The CMDHB Middlemore Hospital includes Kidz First Children's Hospital which provides secondary acute emergency and inpatient care for children aged 0-14 years from the CMDHB geographic area. Ninety-five percent of children resident in the CMDHB area who are admitted with LRI are admitted to Kidz First. Children aged less than 2 years admitted to Kidz First with a primary diagnosis of LRI from 1 August 2007 to 23 December 2007 were eligible. Those admitted with a clinical diagnosis of a lower respiratory infection (bronchiolitis, pneumonia, bronchopneumonia, bronchitis, empyema/lung abscess/TB/pertussis/pneumonitis ) (ICD 10 bronchiolitis, ICD 9466.1, ICD 1010 J 21, pneumonia ICD 948486, 487, ICD 10 J 12-18, J 100 and J110, acute and specified lower respiratory infection ICD 10 J 22, acute bronchitis ICD 9 466, ICD 10 J20, lung abscess and parathorax ICD 9 510, 513, ICD 10 J 85-86, pertussis ICD 9 033, ICD 10 A37) were included. Study procedures—Eligible patients were identified as part of daily screening of admissions by the study nurse and consent was obtained from parents for study participation. Detailed information on household characteristicswas sought in a face to face parent questionnaire which included information on family, maternal age, housing, heating, smoke exposure and number in the house. Number of rooms in the house excluded bathrooms, showers, toilets, laundries, halls, garages and pantries. Socioeconomic deprivation for each child was estimated using the NZDep20066index for area of their residential address. The NZDep2006 index combines 9 variables from the 2006 NZ Census. Individual area scores are then ranked and placed on an ordinal scale from 1 to 10, with decile 10 representing the most deprived 10% of small areas. Children were clinically reviewed daily by the study nurse and the severity of illness documented. Nasopharyngeal aspirates were performed for viral and bacterial testing (data not presented in this paper.7) Relative risk of subgroups for at least one admission was calculated using the CMDHB 2007 birth cohort (data from National Minimum Dataset) as the denominator. In those admitted to hospital with LRI an analysis of risk factors related to total length of hospital stay in the 5 months of the study was performed. A general linear model was fitted with the log of the total length of stay (the sum of all admissions) for a child as the outcome and child's age (at start of study), mothers age, ethnicity of child, deprivation index of home address (coded as 1-8 or 9,10) whether or not premature (<36 weeks), household density (number of rooms in house divided by the number of people in the house), number of smokers in the home and whether or not there was a source of heating in the house included as explanatory variables. Analyses were also run with these variables included individually to ensure associations were not being obscured because of correlation among these explanatory variables. The study was approved by the Northern Regional Ethics Committee (NTX/07/07/059). Results During the study period there were 465 children with 580 admissions to Kidz First wards with LRI. Of these, 394 children (85%) with 508 admissions consented to the study. These admissions were similar to the total group in ethnicity and diagnosis. Table 1. Characteristics of study children admitted to CMDHB facilities with LRI, 1 August 2007-23 December 2007 Variables Cohort N=394 % CMDHB births 2007 n=8833 % Ethnicity Māori Pacific Other 118 221 55 29.9 56.1 14.0 1962 2808 4063 22.2 31.8 46.0 Deprivation 1 (least) to 8 9 10 (most deprivation) 84 124 186 21.3 31.4 47.2 3990 1747 3066 45.2 19.8 34.7 Prematurity Less than 33/40 33/40 to 35+6 36/40 or more 25 33 336 6.3 8.4 85.3 107 276 8309 1.2 3.1 94.1 Table 1 compares the characteristics of the study population with that of the 2007 birth cohort. Relative risk for admission was 4.4 (95%CI 3.2-6.2) for Māori, 5.8 (95%CI 4.4-7.9) for Pacific peoples compared with others, and 3.1 (2.4-3.9) for those living in the most deprived quintile (NZ Dep 9 and 10) compared with those in NZ Dep 1-8 areas. Relative risk of admission was 5.8 (3.8-8.7) for those born at less than 33 weeks gestation and 3.0 (2.1-4.3) for those born at 33<36 weeks gestation compared with those born after 36 weeks. Of the 508 admissions, 323 (64%) had a clinical diagnosis of bronchiolitis, 132 (26%) pneumonia, and 53 (10%) other LRI. Seventy-one (18%) had more than one admission and number of admissions ranged from 1 to 6. Length of stay ranged from 1 to 27 days with a median of 3 days. There was a total of 121 ICU bed days in 21 (5%) patients. The characteristics of households are detailed in Table 2. It is of note that two-thirds of children were exposed to smoke in their home and that two-thirds of families had English as their first language with most others speaking Pacific languages (Table 2). The median number of people in the house was 6, with range 2-31. Median number of adults was 2, median number of children 0-14y was 3, and 100 children (25%) lived in households with ≥7 other people. Thirty-three percent lived in households with four or more children. 224 (57%) had some period of full breastfeeding, 86 (22%) were partially breastfed. Twenty-seven percent lived in houses where caregivers reported never using any form of heating (Table 3). Thirty-nine percent of Pacific families compared with 11% of Māori and 11% of other families had no source of heating. Table 2. Characteristics of households of study children admitted with LRI to CMDHB, 1 August 2007-23 December 2007 Maternal age N % <20 yrs 20-24 yrs >25 yrs 60 117 217 15.2 29.7 55.1 Birth order 1st child 4th or subsequent 125 98 31.7 24.9 First language English Pacific Asian Māori Child in house in daycare No one in house in employment 266 113 8 5 111 44 67.5 28.7 2 1.2 28 11 Access to car during day Yes No 336 58 85.3 14.7 Access to phone Yes No 380 14 96.4 3.6 Smoking exposure Yes No 257 137 65.2 34.8 Table 3. Heating source for households with study children aged <2 years admitted with LRI to Kidz First Children's Hospital, 1 August 2007-23 December 2007 Heating source N % None Electric Bottled gas Wood Mains gas Coal Solar Other 106 210 68 54 10 5 1 3 26.9 53.3 17.3 13.7 2.5 1.3 0.3 0.8 Therapy given during the admission is illustrated in Table 4. Table 4. Therapies received by study children <2 years admitted with LRI to Kidz First Children's Hospital 1 August 2007 to 23 December 2007 by admission Therapies N % Respiratory Oxygen CPAP Ventilation 361 24 2 71.0 4.7 0.0 Fluids NG fluids IV fluids 234 118 46.1 23.2 Medications Ventolin Adrenaline Atrovent 140 34 14 23.2 6.7 2.7 Reflecting severity, 5% of admissions received CPAP and 87% (n=441) received oxygen, salbutamol or supplementary fluids. In 57% (288) of the admissions supplementary fluid was given either via NG tube (46%) and/or as intravenous fluids (23%). In 224 admissions a full blood count was performed, 56 (25%) had an abnormal haemoglobin. A primary care doctor was visited prior to 397 (78%) admissions. Following 390 (77%) admissions some form of follow-up was recommended after discharge, with 271 (53%) referred to the short-term Kidz First home nurse visiting service. In those admitted to hospital the total days in hospital during the study period was associated with prematurity (p=0.005), ethnicity (p=0.02) and age of child (p=0.0002) with those who were premature, or Māori or Pacific ethnicity likely to have a stay approximately 40% longer than respectively those not premature or of European ethnicity. The model also predicted every increase of a month in the child's age to result in a 3% reduction in length of stay (see Table 5). The estimates of the effect sizes differed only marginally when the explanatory variables were analysed individually, with no change in the variables able to be shown to be associated with length of stay. Table 5: Regression coefficient estimates for the log of total length of stay (the sum of all admissions) for a child as the outcome and child's age in days (at start of study), mothers age in years, ethnicity of child, deprivation index of home address (coded as 1-8 or 9,10) whether or not premature (<36 weeks), household density (number of rooms in house divided by the number of people in the house), number of smokers in the home and whether or not there was a source of heating in the house included as explanatory variables.. Source Estimate SE* p value Ethnicity Māori v Pacific