23rd November 2012, Volume 125 Number 1366

Philippa Anderson, Elizabeth Craig, Gary Jackson, Catherine Jackson

In New Zealand, as in other developed countries, there has been increasing interest in healthcare efficiency and the reduction of hospitalisations for conditions amenable to early intervention in primary care. In this context, coding algorithms which can be used to monitor ambulatory sensitive conditions in hospital admissions data, have found favour with those wishing to establish key performance and accountability measures in primary health care.1

In New Zealand, the introduction of an ambulatory sensitive hospitalisation (ASH) target for children (0–4 years) in 20071 was the stimulus for the current project. Previously, ASH rates had been monitored using coding algorithms developed by Jackson and Tobias,2 who grouped potentially avoidable hospitalisations into three broad categories: Preventable Hospitalisations (PH) – by means of population based interventions (e.g. tobacco excise tax); Ambulatory Sensitive Hospitalisations (ASH) – preventable by early and effective treatment in primary care; and Hospitalisations avoidable through Injury Prevention (IP) – preventable via injury prevention measures (e.g. seatbelts).

The coding algorithms used however, were based on tools which had been developed for adult populations, and then updated and adapted for use in the New Zealand context. As a result, they contained many conditions irrelevant to children (e.g. stroke, emphysema), or where preventability was assigned with reference to adult treatment protocols (e.g. urinary tract infections).2

In addition New Zealand has moderate-high child poverty rates,3,4 and hospitalisation rates for many common paediatric conditions have large socioeconomic gradients,5 potentially suggesting that the underlying determinants of health (e.g. housing quality, nutrition, exposure to second hand smoke) are significant drivers of hospitalisations in this age group.6

There was some concern that a health sector target focused solely on access to primary care might have created undue expectations regarding the extent to which primary care could achieve significant reductions in acute paediatric hospitalisations. Such a narrow focus might also divert attention from broader policy measures (e.g. housing quality and affordability, family income support, and childcare), which might potentially achieve greater health gains in this age group.7

With these issues in mind, the project team set out to develop a tool to monitor avoidable hospitalisations in the New Zealand paediatric population. Specifically, the project aimed to:

  • Reframe the current concept of “avoidable” to include policy measures which influenced the socioeconomic determinants of child health. Within this broader framework a sub-set of conditions could then be identified which were amenable to early treatment in primary care. Such an approach aimed to ensure that the role primary care played in preventing hospitalisations would always be seen in the context of broader approaches to address the underlying determinants of child health.
  • Combine this concept of “avoidable” with the clinical expertise of child health professionals to develop a coding algorithm which identified potentially avoidable hospitalisations (PAH) and ambulatory care sensitive conditions (ACSH) in paediatric hospitalisation data, as well any data filters required to ensure that these hospitalisations could be monitored in a consistent manner over time.

Methods

Previous key work to develop coding algorithms for potentially avoidable and ambulatory sensitive conditions has relied heavily on expert consensus.2, 8,9 The exact process by which this has been done is not well described in the literature with most articles stating consensus by expert panel but not elaborating in detail about the process by which consensus was reached. When details of the panel have been given, the numbers have varied with the maximum number of panellists noted to be 17. For this project a hybrid approach (combining Delphi and Nominal approaches) was adopted,10with indicator development progressing as follows:
Developing selection criteria to define avoidable and ambulatory care sensitive conditions—Drawing heavily on the literature linking Government health and social policies to the socioeconomic determinants of health and to child health outcomes,11 12 the authors developed a set of selection criteria to define membership of the potentially avoidable and ambulatory care sensitive categories.
Potentially avoidable hospitalisations (PAH) included those which might potentially be avoided by:
  • Government policies which ensured adequate socioeconomic resources were available to families with children (e.g. income support, childcare, assistance for solo parents returning to workforce).
  • Central and local government policies which ensured that families with children had access to high quality housing and a safe physical environment (e.g. availability, quality and affordability of state and other housing options).
  • Access to timely, appropriate and affordable primary healthcare.
  • The implementation of population based health promotion strategies aimed at improving child health (e.g. adult smoking cessation).
  • Central and local government policies which ensured that the principles of the Treaty of Waitangi (between indigenous Māori and the Crown)13 were taken into account when making resource allocation decisions affecting families.
Two different criteria were used to define the Ambulatory Care Sensitive subset. These were:
  • Early access to primary care could potentially prevent hospitalisation for this condition (e.g. early and appropriate antibiotic treatment for skin infection).
  • If managed appropriately (e.g. early access, adherence to treatment guidelines), the condition could be managed almost entirely in the primary care setting.
Thus by definition, if a condition met the criteria for the Ambulatory Care Sensitive subset, it would also be deemed to be Potentially Avoidable, as a result of it meeting Criterion 3 for PAH.
Data inclusion and exclusion criteriaA number of data inclusion and exclusion criteria were also developed. While some related to unique limitations of the national data collections used, others had wider relevance. These were:
  • Only the primary diagnosis was included for coding purposes to prevent double counting or confusion in coding for admissions with multiple contributing diagnoses.
  • Coding algorithms included those 29 days–14 years but excluded neonates (due to the differing aetiology of neonatal vs. community acquired infectious diseases and the likely lower threshold for hospital admission in the neonatal age group).
  • Age restrictions for some infectious / vaccine preventable diseases (e.g. pertussis was not considered avoidable until >6 months, when the first set of routine vaccinations was complete).
  • All acute (immediate) and semi-acute (occurring within one week of referral) admissions were included. Waiting list admissions were excluded on the basis that elective surgery throughput is often more determined by the supply capacity of a hospital than the burden of need in the community. The only exception was dental admissions, an important cause of avoidable morbidity in the paediatric population. In New Zealand hospitals manage dental admissions variably with some admitting children semi-acutely and others drawing them from a waiting list. Because of this variability all dental admissions were included (irrespective of whether they were acute, semi-acute or drawn from the waiting list).
  • All emergency department (ED) cases were included because of variable paediatric ED assessment procedures around the country.
  • Injuries and poisonings are important causes of avoidable hospitalisations in children. They were not included in the current project however due to variability in the way different EDs upload minor injury cases to the national dataset.
  • Iatrogenic causes of childhood morbidity are considered an important (and under recognised) cause of hospitalisation / prolonged hospitalisation. It was felt however, that such cases were likely underreported (either because they were recorded in secondary diagnostic fields or because of regional inconsistencies in reporting).
Using selection criteria to identify avoidable and ambulatory sensitive conditions—In developing a list of candidate conditions for further scrutiny, the principal diagnoses (ICD-10-AM) for all acute and semi-acute hospitalisations in New Zealand children (aged 1 month–14 years) during 2003-2005 were reviewed and the 40 most frequent conditions identified. To this list a small number of rarer conditions were added. These conditions had been identified by a previous project14 as being important to child health, either because of their severity (e.g. meningococcal disease) or their significant long term costs (e.g. acute rheumatic fever and rheumatic heart disease).
This process resulted in a list of 42 conditions which was then given to a panel of 17 child health professionals (8 Paediatricians, 2 Paediatric Registrars, 2 Public Health Physicians, 2 Public Health Registrars, 1 Paediatric Nurse Specialist, 2 General Practitioners) who were asked to score each condition against the 5 Avoidable and 2 Ambulatory Care Sensitive criteria, using the scale below:
0
Almost totally unavoidable (e.g. <5% could be prevented by this factor)
1
Largely unavoidable (e.g. 5–50% could be prevented by this factor)
2
Partially avoidable (e.g. 51–80% could be prevented by this factor)
3
Largely avoidable (e.g. 81–95% could be prevented by this factor)
4
Almost totally avoidable (e.g. >95% could be prevented by this factor)
In order to assist panel members to consider the role socioeconomic resources (e.g. PAH Criterion 1:Government policies which ensured adequate socioeconomic resources were available to families with children) played in the genesis of each condition, hospitalisation rates by New Zealand Deprivation Index (NZDep2001) were provided.15 The NZDep2001 is a small area index of deprivation which includes 9 variables (reflecting income, employment, communication, transport, support, qualifications, living space and own home) which for most analyses is converted to a decile scale – with decile 1 representing the least deprived 10% of small areas, and decile 10 the most deprived 10% of small areas.15 Panel members were directed to consider the magnitude of the rate ratio (rate NZDep decile 9–10/decile 1–2) for each condition, when considering the impact family resources might have had on the likelihood of hospital admission.
While most participants had little difficulty with the selection criteria, a number felt that the criterion relating to the Treaty of Waitangi should apply equally to all conditions, potentially reducing its utility as a differential scoring tool. Given that the intention of this criterion was to ensure that the Crown’s obligations under the Treaty of Waitangi (i.e. to ensure equity in outcomes for Māori16 were taken into consideration) it was decided that ethnic gradients for each condition would be used to score this criterion. Therefore ethnic specific hospitalisation rates were calculated using the Ministry of Health’s level 1 prioritisation algorithm.17
This hierarchical algorithm allocates children reporting multiple ethnic affiliations to one of five ethnic groups in the following order: Māori, Pacific, Asian, Other, European (e.g. a child identifying with both Māori and Pacific ethnic groups would be counted as Māori).18 For each condition rates for Māori, Pacific and Asian children were then compared to those of European children. Scoring was then assigned on the basis of the rate ratio (RR) as follows: RR≥3.0=Score 4; RR 2.5–2.9=Score 3; RR2.0–2.4=Score 2; RR<2.0= Score 1.
The mean score for each of the five potentially avoidable and two ambulatory care sensitive criteria was then calculated for each condition. If the mean score for any potentially avoidable criterion was ≥2.5 then the condition was considered a cause of potentially avoidable hospitalisation. This cut off was chosen as conceptually it fell half way between a score of 2 (partially avoidable i.e. 51–80% could be prevented by this factor), and a score of 3 (largely avoidable i.e. 81–95% could be prevented by this factor).
Similarly if a mean ambulatory care sensitive criterion score was ≥2.5 the condition was considered a cause of ambulatory care sensitive hospitalisation. Conditions where the highest mean criteria score was between 2.0 and 2.5 were reviewed by a focus group (consisting of 4 medically qualified health professionals) and category assignment was made by consensus, whereas conditions where the highest mean criteria score was <2.0 were automatically excluded.
A draft list of Potentially Avoidable and Ambulatory Care Sensitive conditions was then constructed and given to panel members for their review. While there was general agreement that the lists were appropriate, the inclusion of croup, epilepsy and urinary tract infections (UTIs) in the ambulatory sensitive group was queried by a small number. After further discussion, an age criterion was applied to urinary tract infections (UTIs) (i.e. only UTIs in children >4 years were considered ambulatory care sensitive).
Fleiss’s Kappa was used to calculate inter-rater agreement between the panel members who scored each of the conditions against the various criteria. The irr package of the R-project statistical software, was used to undertake these analyses.
Ethics approval was not required.

Results

Table 1 summarises the mean scores the 42 conditions on the candidate list received against each of the five Potentially Avoidable and two Ambulatory Care Sensitive Criteria. Similarly, Table 2 and Table 3 summarise the final lists of Potentially Avoidable and Ambulatory Care Sensitive conditions respectively.

Table 1. Mean scores for each condition against the 5 avoidable and 2 ambulatory sensitive criteria

Condition

Avoidable Under Wider Definition

Avoidable by Appropriate Access to Primary Care

Social Policy

Housing / Physical Environment

Access to Primary Care

Health Promotion

Ethnic Disparities

Early Access

Managed in Primary Care

Abdominal/pelvic pain






Acute appendicitis







Acute URTI NOS



Acute bronchiolitis





Acute rheumatic fever



Acute upper respiratory infections



Apnoea/ breath holding







Asthma


Bronchiectasis



Bacterial meningitis





Bacterial pneumonia


Chemotherapy







Coagulation defects







Constipation




Chronic rheumatic fever (rheumatic heart disease)



Croup




Cystic fibrosis







Dental (dental caries, pulp, periodontal)




Dermatitis/eczema



Epilepsy







Failure to thrive





Febrile convulsions




Gastroenteritis-bacterial/protozoal



Gastroenteritis-other



Gastro oesophageal reflux





Immune disorders






Inguinal hernia







Meningococcal disease





Mental health and behaviour disorders






Neoplasms







Nutritional deficiency



Otitis media


Osteomyelitis






Pertussis




Skin infection

Type 1 diabetes







Tuberculosis





Urinary tract infection




Vaccine preventable diseases



Viral pneumonia






Viral meningitis






Viral infection of unspecified site



Note: For each condition, panel members awarded a score of 0-4 for each criterion; ■ = mean score ≥ 2.5, ▲= mean score 2-2.49; ARF= acute rheumatic fever; CRF= chronic rheumatic fever; URTI= upper respiratory tract infection; VPD= Vaccine preventable disease; NOS= not otherwise specified. All conditions that scored ≥ 2.5 were considered avoidable. Conditions where the highest mean criteria score was between 2.0 and 2.49 were reviewed by a focus group and category assignment was made by consensus, whereas conditions where the highest mean criteria score was <2.0 were automatically excluded.
Table 2. Final list of potentially avoidable hospitalisations
Condition
ICD-10-AM Code
Potentially Avoidable Hospitalisations
Acute bronchiolitis
J21
Acute rheumatic fever
I00-I02
Acute upper respiratory tract infection excluding croup
J00-J03, J06
Asthma
J45,J46
Bronchiectasis
J47
Bacterial meningitis*
G00,G01
Bacterial/ Unspecified pneumonia
J13-J16, J18
Constipation
K590
Chronic rheumatic heart disease
I05-I09
Croup, acute laryngitis, tracheitis
J04 J050
Dental (dental caries, pulp, periodontal)
K02,K04,K05
Dermatitis/eczema
L20-L30
Febrile convulsions
R560
Gastroenteritis
A00-A09,R11, K529
Gastro oesophageal reflux
K21
Meningococcal disease
A39
Nutritional deficiency
E40-E64, D50-D53
Otitis media
H65-H67
Osteomyelitis
M86
Vaccine preventable diseases
tetanus neonatorum congenital rubella
tetanus, diphtheria, pertussis, polio, hepatitis B,
measles, rubella, mumps
P350,A33,A34
A35,A36, A37,A80, B16,B180,B181
B05,B06,B26, M014
Skin infection
L00-L05,L08,L980,J340,H010,H000
Tuberculosis
A15-A19
Urinary tract infection ≥ 5 years
N10, N12,N300,N390,N136,309
Viral pneumonia
J12, J100,J110
Viral / other / unspecified meningitis
A87,G02,G03
Viral infection of unspecified site
B34
Not Potentially Avoidable Hospitalisations
Abdominal/pelvic pain
R10
Acute appendicitis
K35
Chemotherapy
Z511
Coagulation defects
D65-D69
Cystic fibrosis
E84
Epilepsy / Status epilepticus
G40,G41
Immune disorders
D80-D89
Inguinal hernia
K40
Neoplasm- malignant or non malignant
C00-D48
Type 1 diabetes
E10
Urinary tract infection < 5 years
N10, N12,N300,N390,N136,309

*Note: Meningococcal meningitis included under meningococcal disease.

Table 3. Final list of ambulatory care sensitive hospitalisations
Condition
ICD10-AM code
Ambulatory Care Sensitive Conditions
Acute rheumatic fever
I00-I02
Acute upper respiratory tract infections excluding croup
J00-J03, J06
Asthma
J45,J46
Bacterial/Unspecified Pneumonia
J13-J16, J18
Bronchiectasis
J47
Constipation
K590
Chronic rheumatic heart disease
I05-I09
Dental (dental caries, pulp, periodontal)
K02,K04,K05
Dermatitis/eczema
L20-L30
Gastroenteritis
A02-A09,R11, K529
Gastro oesophageal reflux
K21
Nutritional deficiency
D50-D53,E40-E64
Vaccine preventable diseases
tetanus neonatorum congenital rubella
> 6 months: tetanus, diphtheria, pertussis, polio, hepatitis B 

>16 months: measles, rubella, mumps

P350,A33,A34
>6months: A35,A36, A37,A80, B16,B180,B181
>16months: B05,B06,B26, M014
Otitis media
H65-H67
Skin infection
L00-L04,L08,L980,J340,H010,H000
Urinary tract infection ≥ 5 years
N10,N12,N136,N300,N309,N390
Non Ambulatory Care Sensitive Conditions
Abdominal/pelvic pain
R10
Acute appendicitis
K35
Acute bronchiolitis
J21
Bacterial meningitis*
G00,G01
Chemotherapy
Z511
Coagulation defects
D65-D69
Croup, acute laryngitis, tracheitis
J04, J050
Cystic fibrosis
E84
Epilepsy/ Status Epilepticus
G40,G41
Febrile convulsions
R560
Inguinal hernia
K40
Immune disorders
D80-D89
Meningococcal disease
A39
Neoplasm—malignant or non malignant
C00-D48
Osteomyelitis
M86
Type 1 diabetes
E10
Tuberculosis
A15-A19
Urinary tract infection <5 years
N10,N12,N300,N390,N136,N309
Vaccine preventable diseases under relevant age cut offs
≤6months: A35,A36, A37,A80, B16,B180,B181
≤16months: B05,B06,B26, M014
Viral pneumonia
J12, J100,J110
Viral / other / unspecified meningitis
A87,G02,G03
Viral infection of unspecified site
B34

*Note: Meningococcal meningitis included under meningococcal disease.

In general, Potentially Avoidable Hospitalisations tended to be viewed as those arising from conditions of an infectious or respiratory nature, while hospitalisations for chronic medical conditions (e.g. cancer, diabetes) or surgical problems (e.g. appendicitis) were viewed as non-avoidable. While a similar pattern was seen for Ambulatory Care Sensitive Hospitalisations, infectious diseases with a predominantly viral aetiology (e.g. bronchiolitis, viral pneumonia) were more likely to be viewed as non-ambulatory care sensitive (i.e. early treatment in primary care was thought unlikely to change their course significantly).

In terms of inter-rater agreement, analysis of scoring using the five Potentially Avoidable and two Ambulatory Sensitive Criteria suggested only slight agreement between panel members. Because of the large number of missing scores for two raters, they were omitted from analyses. Kappa coefficients (Fleiss) for the 5 Potentially Avoidable Criteria ranged from 0.140 for ‘Primary Care’ to 0.150 for the ‘Health Promotion’ criterion. Kappa coefficients for ‘Ambulatory Care’ ranged from 0.095 for ‘Early Access’, to 0.170 for the ‘Completely Managed in Primary Care’ criterion.

Discussion

This is the first time that Potentially Avoidable and Ambulatory Care Sensitive Hospitalisation indicators have been developed specifically for the New Zealand paediatric population. The tools developed represent a significant improvement on those used previously in New Zealand 2which contained diagnoses inappropriate for children or had failed to take into account age related differences in the aetiology or management of common conditions (e.g. UTIs).

In addition, the broadening of the concept of avoidable to include the policies which shape the underling determinants of health serves to place the role of primary care in the prevention of acute paediatric hospitalisations within a broader context. While the tools developed are a considerable improvement on those used to date, the use of diagnostic coding algorithms to monitor ambulatory care sensitive hospitalisations and by inference, the performance of primary care, remains problematic for a number of reasons.

Firstly, the extent to which ambulatory care sensitive hospitalisations are actually avoidable remains unclear. One survey of 554 children hospitalised for ambulatory sensitive conditions in the USA noted that only 25% of parents, 29% of primary health care physicians and 32% of hospital doctors felt that their child’s / patient’s admission could have been avoided given improved access to primary care, better attention to preventative medication, or avoidance of known triggers.19 It is also notable that the level of agreement between parents and primary health care physicians as to whether their child’s / patients admission might have been avoidable (kappa 0.31), was not much higher than the level of agreement seen between panel members when scoring avoidability in the current study.19

Secondly, in New Zealand primary diagnoses are assigned at the time of hospital discharge after all relevant investigations have been undertaken. Therefore while at first glance it may appear that 100% of admissions for acute upper respiratory infections should be avoidable, given early access to primary care, in reality such a diagnosis may be one of exclusion, arrived at only after more serious causes of illness have been ruled out by investigations unavailable in primary care (e.g. lumbar puncture). Therefore any initiatives aimed at reducing such seemingly trivial hospitalisations must be carefully weighed up in order to ensure that patient safety is not compromised in the quest for health service efficiency.

Thirdly, the predominance of acute infectious and respiratory diseases seen in Table 3 is in sharp contrast to the ambulatory care sensitive conditions identified by Jackson and Tobias2 for older adults where chronic conditions (e.g. angina, hypertension) predominate. Such differences potentially suggest that the window of opportunity available for successful primary care intervention in children (e.g. acute respiratory infections = hours-days) is much briefer than for adults (e.g. antihypertensives to prevent ischemic heart disease= months-years), and as a consequence, a much greater emphasis needs to be placed on providing access to immediate and afterhours primary care.

Fourthly injuries, poisonings and iatrogenic causes of childhood morbidity, which are responsible for a large number of childhood hospitalisations, were not included in the tool as described above. As data quality improves the development of injury and iatrogenic subsets are recommended.

Finally, the complex interrelationships between socioeconomic factors, access to primary care, and hospitalisations for ambulatory sensitive conditions are difficult to untangle. During office hours the majority of children under the age of 6 years in New Zealand receive free primary health.20 However after hours care is invariably subject to co-payments, and appointments are not always available. Further primary care for older family members (including children > 6years) usually attracts user charges, with parents often being reluctant to seek free care for their children from practices where they owe debts for older family members.21 In contrast, secondary care (including attendance at Emergency Departments) remains free for all age groups, thereby creating incentives for families with restricted economic resources to bypass primary care, particularly in the after-hours context.

The inconsistent uploading of emergency department (ED) cases to the hospital admission dataset further complicates this issue by making it difficult to consistently remove “walk in” ED cases from any analysis of avoidable hospitalisations. In New Zealand, an event is recorded as a hospital admission if treatment time exceeds 3 hours, irrespective of the department in which the patient is assessed. While for adults, some analysts exclude from analysis all patients admitted to an ED and discharged alive the same day, for paediatric cases such an approach is problematic. This is because large urban centres tend to assess children presenting acutely in specialist paediatric EDs, while many smaller centres send similar cases to the paediatric ward / assessment unit (where they are assigned an inpatient code).

The subsequent filtering out of ED cases thus disproportionately discounts the work of larger urban centres, who manage much of their patient throughput via specialist paediatric EDs. The inclusion of emergency day cases remains controversial however as a number of these children are likely to be from families attempting to access free primary health care via secondary or tertiary hospital services.

In addition, New Zealand has moderate-high child poverty rates,3,4 and high rates of household crowding 21,22 and exposure to second hand cigarette smoke,23 each of which are significant drivers for acute infectious and respiratory diseases in children. Therefore, in the context of a funding model which incentivises families to seek after hours care from hospital EDs, an infectious and respiratory disease burden which is being driven by a complex web of socioeconomic causality, and a narrow window for effective intervention, the extent to which ‘ambulatory care sensitive’ hospitalisations can or should be used to monitor primary care performance remains debatable.

The large social gradients evident for many of these conditions also suggests that effective government policies implemented by agencies which sit outside of the health sector (e.g. housing, social welfare) may potentially result in large reductions in childhood hospitalisations for potentially avoidable conditions).

Conclusions

This paper describes the development of a tool to measure Potentially Avoidable and Ambulatory Care Sensitive Hospitalisations in the New Zealand paediatric population, using a methodology which includes the broader determinants of health in the conceptualisation of what is Avoidable. While policies which ensure early access to effective primary health care are crucial for optimal child health outcomes, the role Government social (e.g. welfare entitlements) and other (e.g. housing, early childhood education) policies play in shaping the underlying determents of health are likely to be as important, in reducing the large burden of avoidable morbidity currently experienced by New Zealand children.

What is already known?

Tools which measure avoidable morbidity have been developed to monitor health services performance

The current tools do not capture the role of socioeconomic determinants on health outcomes for children

What this study adds

The PAH tool incorporates the broader determinants of health in order to provide information about the potential health gain possible for the paediatric population.

The ACSH tool provides a more accurate reflection of the role of primary care’s ability to impact on hospitalisation rates than previous indicators used for the paediatric population in New Zealand.

What are the policy implications?

The PAH tool draws attention to the sectors outside the health system that play an important role in determining health outcomes for children and therefore have the potential to influence government policy in these areas

Summary

There have been a number of ‘tools” or indicators used to try and capture the number of hospitalisations that could be prevented if people were able to access primary care in a timely way and receive appropriate care. These have been referred to as Ambulatory Care Hospitalisations (ASH). In New Zealand there had not been a specific list of ASH conditions for children that reflect Primary Care’s ability to influence hospitalisation rates for children. This paper describes how a group of ASH conditions was identified for children. In addition it has been recognised that factors outside the health system (government social and other policies) have a large impact on heath outcomes for children. The paper describes the methodology for identifying this group of conditions named Potentially Avoidable Hospitalisations.

Abstract

In New Zealand there has been increasing interest in reducing avoidable hospitalisations, particularly from conditions treatable in primary care. To date avoidable hospitalisations in children have been monitored using adult tools which contain many conditions irrelevant to children. Further, New Zealand has large socioeconomic gradients in hospitalisations for many paediatric conditions, suggesting that the social determinants of health also heavily influence avoidable hospitalisations in this age group.

Aim

(1) To develop a tool to monitor potentially avoidable hospitalisations in New Zealand children which includes the socioeconomic determinants of health within the conceptualisation of “avoidable”; and (2) Within this broader framework, to identify a sub-set of conditions which are amenable to intervention in primary care.

Method

Five selection criteria were developed to define Potentially Avoidable Hospitalisations (PAH), and a further two criteria were used to define a subset of Ambulatory Care Sensitive Hospitalisations (ACSH). The principal diagnoses for all acute hospitalisations in New Zealand children (1 month–14 years) during 2003-2005 were then reviewed, and a list of 42 conditions created. This list was sent to 17 health professionals with experience in child health, who were asked to score each condition against the 5 PAH and 2 ACSH criteria.

Results

Twenty-six conditions contributing to PAH were identified, along with 18 contributing to ACSH. PAH tended to be infectious or respiratory in nature, with hospitalisations for chronic medical conditions or surgical problems being viewed as non-avoidable. While a similar pattern was seen for ACSH, viral infections were viewed as non-ambulatory care sensitive.

Conclusion

While the tools developed are a considerable improvement on those used to date, the use of diagnostic coding algorithms to monitor ACSH and by inference, the performance of primary care, remains problematic for a number of reasons. Nevertheless, the broadening of PAH to encompass the wider determinants of health, serves to highlight the role Government social and other policies might play in reducing the large burden of avoidable morbidity currently being experienced in this age group.

Author Information

Philippa Anderson, Public Health Physician, Counties Manukau District Health Board, Auckland; Elizabeth Craig, Public Health Physician, Director of the New Zealand Child and Youth Epidemiology Service. Department of Women’s and Children’s Health at the University of Otago’s Dunedin School of Medicine, Dunedin; Gary Jackson, Public Health Physician, Health Partners Consulting Group Limited, Auckland; Catherine Jackson, Public Health Registrar, Auckland Regional Public Health Service, Auckland

Acknowledgements

The work of PA was supported by a training grant from the New Zealand Population Health Charitable Trust (this work received no other external funding).The authors also thank Dr Simon Thornley for kindly providing the Kappa analysis for this study, and Dr Martin Tobias, who provided expert advice during the course of PAH and ACSH tool development and editorial advice on the final drafts of this manuscript.

Correspondence

Philippa Anderson, 19 Lambie Drive, South Auckland Mail Centre 2104, New Zealand.

Correspondence Email

Philippa.Anderson@middlemore.co.nz

Competing Interests

No financial interests are involved. This work and any views expressed are solely those of the authors, and not of their employing agencies. No editorial control came from either organisation.

References

  1. Ministry of Health. Health targets: Moving Towards Healthier Futures 2007/2008. Wellington: Ministry of Health, 2007:1-39.
  2. Jackson G, Tobias M. Potentially Avoidable Hospitalisations in New Zealand, 1989-98. Aust N Z J Public Health 2001;25(3):212-21.
  3. The Ministry of Social Development. The Social Report. Wellington: The Ministry of Social Development., 2006.
  4. The Asthma and Respiratory Foundation of New Zealand. Trying to Catch Our Breath. In: Asher MI, Byrnes CA, editors. Wellington: The Asthma and Respiratory Foundation of New Zealand, 2006.
  5. Craig E, Anderson P, Jackson C. The Health Status of Children and Young People in Counties Manukau: The New Zealand Child and Youth Epidemiology Service, 2008.
  6. The National Advisory Committee on Health and Disability. The Social, Cultural and Economic Determinants of Health in New Zealand: Action to Improve Health. Wellington: The National Advisory Committee on Health and Disability, 1998.
  7. National Health Committee. The Social, Cultural and Economic Determinants of Health in New Zealand: Action to Improve Health. Wellington: National Advisory Committee on Health and Disability, 1998.
  8. Billings J, Teicholz N. Uninsured patients in District of Columbia Hospitals. Health Aff (Millwood) 1990;9:158-65.
  9. Begley CE, Slater CH, Engel MJ, Reynolds TF. Avoidable hospitalizations and socio-economic status in Galveston County, Texas. J Community Health 1994;19(5):377-87.
  10. Hutchings A, Raine R, Sanderson C, Black N. A comparison of formal consensus methods used for developing clinical guidelines. Journal of Health Services & Research Policy 2006;11(4):218-24.
  11. Public Health Advisory Committee. The Health of People and Communities. A way Forward: Public policy and the economic determinants of health. Wellington: The Public Health Advisory Committee (subcommittee of the National Health Committee), 2004.
  12. King J. Economic Determinants of Health: A Report to The Public Health Advisory Committee, Unpublished, 2003.
  13. Ministry of Health. The New Zealand Health Strategy. Wellington: Ministry of Health, 2004.
  14. Craig E, Jackson C, Han D, NZCYES Steering Committee. Monitoring the Health of New Zealand Children and Young People: Indicator Handbook. Auckland: Paediatric Society of New Zealand & New Zealand Child and Youth Epidemiology Service, 2007.
  15. Salmond C, Crampton P. NZDep2001 Index of Deprivation. Wellington: Department of Public Health, Wellington School of Medicine and Health Sciences, 2002:1-57.
  16. Cunningham C, Kiro, C. Rapuhia mo te Hauroa Maori in Health and Public Policy in New Zealand. In: Davis P, Ashton T, editors. Auckland: Oxford University Press, 2001:62-79.
  17. Ministry of Health. Monitoring Ethnic Inequalities in Health. Public health Intelligence Occasional Bulletin No.4. Wellington: Ministry of Health, 2001.
  18. Ministry of Health. Ethnicity Data Protocols for the Health and Disability Sector. Wellington: Ministry of Health, 2004.
  19. Flores G, Abreu M, Chaisson CE, Sun D. Keeping children out of hospitals: parents' and physicians' perspectives on how pediatric hospitalizations for ambulatory care-sensitive conditions can be avoided. Pediatrics 2003;112(5):1021-30.
  20. Ministry of Health. A Portrait of Health-Key Results of the 2006/07 New Zealand Health Survey: Ministry of Health, 2008.
  21. Jensen J, Krishnan V, Hodgson R, Sathiyandra S, Templeton R, Jones D, et al. New Zealand Living Standards 2004. Wellington: Ministry of Social development, 2006.
  22. Baker M, McNicholas A, Garrett N, et al. Household crowding a major risk factor for epidemic meningococcal disease in Auckland children. Pediatr Infect Dis J 2000;19(10):983-90.
  23. Craig E, McDonald G, Reddington A, Wicken A. The Determinants of Health for Children and Young People in Counties Manukau. Dunedin: New Zealand Child and Youth Epidemiology Service. Department of Women's and Child Health. University of Otago, 2009:147-55.