This study has uniquely described patterns of health service utilisation in the total infant population of two of the biggest remote communities in Australia's Northern Territory. We have documented extremely high rates of health service utilisation at the primary and referral level, commencing from birth and continuing throughout the first year. Remote-dwelling Aboriginal infants access health care frequently for both routine and acute care, despite the multiple barriers to care outlined by others [15, 16].
There were a total of 11,224 presentations to the HCs for the three years of data collection. For each community, this translates to an average of 7.65 infant presentations per day (based on 249 working days in the year). The implications of this for remote workforce planning are important given that most HC presentations were for new, acute problems. The severity and complexity of many presentations in these HCs can require multiple staff to provide numerous hours of acute care to an individual infant, particularly when the infant needs emergency air evacuation to hospital. Cultural and linguistic barriers as well as staffing shortages, a lack of nurses with child health skills and qualifications and rapid turnover resulting in repeated training of new staff  compound this workload in remote health services. The organisation and delivery of infant health services in remote northern Australia varies across HCs. Some HCs have specific days for routine health checks by designated staff, with the acute care delivered by other staff as needed. Other communities have designated staff that delivers both routine and acute care any time that the infant presents to the HC.
Service provision is dependent upon HC funding (staffed for a 5 day week, minimal weekend service and on call service afterhours; not 24/7 service provision), availability of staff (relief not always provided for holidays or educational leave), callouts the previous night, staff skill mix and community size. Current staffing levels for infant and child health services in remote communities are not determined by their burden of disease or service usage and are insufficient to meet the needs of the young population, thus affecting the quality of care .
We distinguished acute care episodes from routine care at the HC. This has not been previously investigated among the remote dwelling Aboriginal infant population in the NT. We identified respiratory, skin and gastrointestinal symptoms as the leading new problems seen at the HC. Others have shown similarly high presentation rates primarily for infectious diseases in remote HCs . High rates of primary health service utilisation have also been identified among suburban Victorian infants, however the bulk of the visits were unrelated to acute illness unlike our findings .
A third of presentations were for routine health checks and other non-acute interventions. Community based workers, Strong Women Workers and AHWs are ideally situated to provide much of this preventive care and health education in a culturally safe framework and potentially reduce the workload for clinical staff busy attending to the burden of acute illness, although this is not currently occurring in many remote settings.
Poor basic living conditions contribute to the burden of disease . However, in an era when the nation is focused on closing the gap in under 5 mortality and health outcomes, providing better care for infants in their first year of life is a critical issue that must be targeted. Health services should be designed to provide high quality health care for infants as well as preventative education and effective interventions for known contributors to poor infant health outcomes such as maternal and household smoking. Ideally this should commence early in the antenatal period.
Several approaches to improving health service effectiveness are being introduced across remote communities including the Healthy Under Five Kids program, designated child and family health nurse positions, and the expansion of family support workers. These programs are in their implementation phase and have not been funded to be rigorously evaluated.
We identified a high rate of hospitalisation. One third of infants were admitted to the NNU following birth. This is double the admission rate for non-Aboriginal infants in the rest of Australia . More than half of the infants admitted were born preterm. The total preterm birth rate was 6% higher in these communities compared with the preterm birth rate among other Aboriginal babies in the NT . Problems with the accurate estimation of gestational age due to poor maternal recall of menstrual period dates and uptake of early ultrasound, are well described in the Australian Aboriginal population [22–24]. We identified 8/42 premature LBW whose true gestation we could calculate based on 1st trimester ultrasound. Only one case of misclassification as premature occurred.
Excluding NNU admissions, 47% of infants had at least one hospital admission before they turn one. The high rates of admissions for respiratory infection identified in our study concur with other NT studies [25, 26].
Despite the large number of visits audited, the retrospective nature of this study limits causal inference and a number of infant records were unavailable for review. It seems likely that these few records were missing completely at random so the impact on inference is likely to be minimal. Data linkage between primary HC and hospital records was complicated by infants with multiple first and surnames and addresses; some misidentification of infants may have occurred. Finally, given the mobility of Aboriginal populations in the NT , infants may have presented for care at other health services or have been admitted to a hospital other than the regional hospital reviewed in this study, in which case our results would only underestimate service utilisation.