View Article PDF

In the context of the current severe acute respiratory syndrome coronavirus 2019 (COVID-19) pandemic, lack of capacity and resources for acute healthcare has been recognised internationally.[[1,2]] Many acute healthcare systems operate near, at or over capacity, with emergency department (ED) crowding and access block (admission delays for acute inpatients) a widely recognised problem worldwide. ED crowding driven by access block impairs quality of care and is a major cause of morbidity and mortality for acute patients.[[3]] It is unclear whether New Zealand’s EDs are prepared for a ‘slow moving mass casualty incident’, such as has been seen in countries like the USA over the past 18 months.[[4]] Another unknown is whether EDs in New Zealand have comparable staffing and capacity, so some regions may be more able to cope with such disasters than others. To facilitate understanding and interpretation of international comparisons of acute care, the International Federation for Emergency Medicine (IFEM) recently developed a template for uniform reporting of ED structure, staffing, governance and performance.[[5]] This survey includes the performance measures time to assessment, which is related to patient satisfaction and experience, and ED length of stay (LOS), which is related to patient outcomes including mortality.[[6,7]]

The primary aim of this study was to document the current state of EDs in New Zealand according to the IFEM template and to look for opportunities to reduce variation nationally. The secondary aim was to explore whether any of the structural and staffing variables were associated with the process and outcome measures.

Methods

This was a retrospective cross-sectional survey of EDs in New Zealand for the calendar year 2018.

Site selection

All 25 EDs in New Zealand that report data to the Ministry of Health (MOH) for the SSED target were eligible to participate. Participants were invited to contribute data via email to the ED clinical director. Participation was voluntary and not incentivised.

Survey instrument

The survey questions were those suggested by the IFEM.[[5]] The template has sections based on ED structure, staffing and governance, attending population, processes and outcomes. In acknowledgement that ED processes and outcomes depend on the bed-state of the hospital, the template also includes a section on the number of hospital beds for the population served by the hospital.[[5]]

Data collection and checking

An electronic link to an online survey managed by the MOH was provided, and on request a Microsoft Word™ document was also available for manual data entry. A manual with data definitions was provided with the survey instruments. The survey opened for responses on 22 May 2019 and closed on 22 July 2019. Survey questions are shown in the supplementary material. Where data were incomplete or incongruent, follow-up emails were sent with a request to complete or check the validity of the data. Data on hospital bed numbers were also cross checked with individual district health board (DHB) websites and the MOH website for each DHB. Where individual hospital populations could not be determined, the DHB population and total DHB hospital bed numbers were used in the calculation of beds/1,000 population. The correlation between the bed numbers reported from survey respondents and the MOH website was r=0.99, with the MOH website estimating 5.2% more beds than the survey respondents.

Data analysis

Data were downloaded and entered into a spreadsheet (Microsoft Excel™ 2016). Analysis was descriptive, using proportion with 95% confidence interval (95%CI), median with interquartile range (IQR) or mean with standard deviation.

The relationship between variables was assessed visually using scatter plots. Where appropriate, correlations were performed using SPSSv22 (IBM corporation, Armonk, NY, USA). The distribution of each variable was assessed visually using histograms and Q-Q plots and statistically with the Kolmogorov–Smirnoff test. Pearson correlation was used for normally distributed variables; otherwise Spearman’s correlation was used. Statistical significance was adjusted for multiple comparisons using the Bonferroni calculation (alpha/number of comparisons), so statistical significance was set at p<0.001. A sample size calculation was not performed as this was an exploratory survey with a maximum sample size of 25.

Ethics

As a survey of administrative data, this study was out of scope for the New Zealand Health and Disability Ethics Committees.

Results

Responses were received from 18/26 hospitals (69%), which together were representative of smaller regional hospitals through to major urban tertiary hospitals. The number of hospital beds ranged from 43 to 1,005, and hospital beds per 1,000 population ranged from 1.5 to 2.3. The number of presentations ranged from fewer than 20,000 per year to more than 100,000, with the presentations per 1,000 DHB population ranging from 163 to 352 (Table 2).

Governance, staffing and structure

Table 1 shows the variation in size, structure and staffing of EDs. The number of beds ranged from 12 with no ED short stay unit to 57 with a 23-bed short stay unit. Staffing also varied: 105 to 259 nursing hours per 100 patient visits; 0 to 28 advanced care (nurse) practitioner hours per 100 patient visits; and 68.7 to 146 physician hours per 100 patient visits. Only one ED reported direct admitting rights, and one reported having an emergency medicine specialist in the ED 24/7.

Table 1: Structure, staffing and governance. View Table 1.

Population presenting to ED

Table 2 shows the variation in patient age and acuity of ED presentations to the different departments. Excluding Starship Children’s Hospital and Auckland City Hospital, which see only children or adults respectively, the proportion of children aged 0–5 ranged from 4.2% to 19.4%. The number of adults over 75 years ranged from 9.6% to 19.3%. There was also a wide variation in the proportion of ambulance arrivals (7.2% to 33.3%). Triage acuity also varied. One marked difference was in the higher proportion of resuscitation cases that presented to Auckland City Hospital (more than 4%) when all other departments saw <1% (Table 2).

Table 2: Presenting population. View Table 2.

Processes and outcomes

Table 3 shows ED process times. The wait to be seen ranged from median of 13 (IQR 6, 37) to 92 (IQR 41, 162) minutes. The ED LOS varied from a median of 135 (IQR 76, 222) to 288 (IQR 177, 517) minutes. With respect to outcomes, patients not waiting to be seen ranged from 1.0% to 8.0%, whereas admission rates ranged from 9.6 to 40.0%. There were few deaths in the ED (<0.01% to 0.14%). The proportion of patients who re-presented to the ED within 72 hours and were subsequently admitted ranged from 0.6% to 5.1%, and all but two sites reported proportions less than 2%.

Table 3: Processes and outcomes. View Table 3.

Relationship of structure, workload and staffing to processes and outcomes

Structural measures

Examination of the scatterplots showed no linear relationship between either ED spaces per 1,000 presentations or hospital beds per 1,000 DHB population for any process or outcome (see the supplementary material).

Workload measures

The scatterplots suggested a there may be a relationship between the proportion of ambulance presentations and most processes and outcomes. There were some moderate correlations:

  • time to assessment r=0.590, p=0.010
  • ED LOS r=0.593, p=0.012
  • did not wait (DNW) r=0.566, p=014
  • deaths in ED r=0.513, p=0.029

Although the proportion of admissions was uncorrelated to outcomes (scatterplots shown in the supplementary material), the proportion of admissions was moderately correlated to ED processes:

  • time to assessment r=0.494, p=0.037
  • ED LOS r=0.609, p=0.009

The proportion of high acuity presentations was associated with the proportion who died in ED (r=0.564, p=0.015) but no other processes or outcomes.

Staff measures

The scatterplots showed no linear relationship between nursing hours or advanced care practitioners per 100 patients and any of the process or outcome measures (see the supplementary material). Doctor hours per 100 patients was negatively correlated to time to assessment (r=-0.617, p=0.006) and DNW (r=-0.515, p=0.029), although not to other processes and outcomes (scatterplots shown in the supplementary material).

Combinations of measures

Exploratory analysis looking at different combinations of structural, workload and staffing measures found stronger correlations with process and outcome measures than with the individual variables:

  • For time to assessment, the strongest correlation was seen when sites had high ambulance presentations, high admissions to hospital and low doctor hours: r=0.728, p=0.001.
  • For ED LOS, the strongest correlation was with high ambulance presentations, high admission rates, fewer beds per 1,000 DHB population and low doctor hours: r=0.759, p<0.001.
  • For DNW, the strongest correlation was with high ambulance presentations and fewer ED treatment spaces per 100,000 DHB population: r=0.619, p=0.006.
  • For deaths in the ED, the strongest correlation was for high ambulance presentations, fewer ED treatment spaces per 100,000 DHB population and low doctor hours: r=0.649, p=0.004.

Figure 1 shows the scatterplots for these associations and the full table of all correlations tested is provided in the supplementary material.

The scatterplots showed that the proportion of patients re-presenting and being re-admitted within 72 hours was not linearly related to any structure, workload or staffing measure, or any combination of these measures, as most sites had fewer than 2% re-presentations requiring admission, with two notable outliers heavily influencing the observed correlations.

Discussion

The survey found marked variation in the presenting population, structure, staffing and governance of New Zealand EDs. The reasons for the variation are likely to be multifactorial and may be inter-related. Some of the variation is most likely due to the locations of hospitals and the availability of alternative care. For example, the populations presenting to smaller rural or regional hospitals that have limited availability of after-hours primary or urgent care was more likely to be lower acuity, since the only available option for care is the ED. This was different in major centres, where urgent care clinics deal with the vast majority of minor injury and medical complaints. This was reflected in the lower presentation rates, higher admission to hospital rates and corresponding longer lengths of stay observed in larger urban tertiary hospitals compared to rural and regional centres. A notable exception among tertiary urban referral hospitals was the sole specialist paediatric hospital, which had fewer patients, lower triage acuity, higher staffing, shorter times to assessment, fewer admissions and few patients leaving prior to assessment. In contrast, the sole adult hospital had a high proportion of triage 1 cases, which reflected its role as the regional trauma, stroke and 24-hour interventional cardiology centre.

There was marked variation in the number of bed spaces per 1,000 presentations, ranging from under 0.6 to more than 1.0. Sites with fewer ED beds per 1,000 presentations were likely to be more fully occupied more of the time and be less able to cope with surges in demand, such as has been seen internationally during the COVID-19 pandemic. Similarly, hospital bed numbers per 1,000 population varied considerably. The average of 1.9 was significantly lower than 2.6 of beds per 1,000 population reported for New Zealand by the Organisation for Economic Cooperation and Development (OECD).[[8]] In the context of a national disaster or pandemic, those hospitals with lower bed number per head of population will less likely cope with surges in demand. In sites with few ED spaces and very low hospital bed numbers per 1,000, there is an opportunity for infrastructure investment to redress this deficiency.

There was wide variation in nursing hours per 100 presentations. No clear pattern with respect to number of presentations was observed. Physician plus advanced care provider ratios per 100 patients varied less, except for the two busiest EDs, which had had the lowest clinician-to-patient ratios. A notable exception was the specialist paediatric hospital, which had much higher staffing ratios than other sites.

Most EDs had a short stay unit run by an emergency medicine specialist, which allowed for longer periods of ED care and prevented unnecessary hospital admissions. The use of ED short stay units increased considerably in the last decade, partially in response to the SSED health target.[[9]]

The reporting measures used in isolation were not associated strongly with the observed variation in performance. Only the proportion of ambulance presentations seemed possibly related to most ED process and outcome measures. The proportion of admissions was also related to ED processes, although the observed relationships were not strong. Unsurprisingly, a higher proportion of ambulance arrivals was associated with more deaths in the ED: patients who arrive by ambulance are typically more unwell. The impact of ambulance arrivals on ED process measures may have reflected the extra time required to investigate and treat these especially unwell patients. It was also recognised that the efficiency of admission processes within the hospital has an influence on ED LOS for admitted patients.[[10]]

When structural, staffing and workload measures were combined, there were clearer relationships between the processes and outcomes of care. When sites with higher proportions of ambulance arrivals and higher admission rates also had fewer doctors, less space or fewer hospital beds, then there were prolonged times to assessment, longer ED LOS, more patients leaving without being seen and more deaths in the ED. Investment in infrastructure and staffing may be required at these sites to improve processes and outcomes of care and reduce variation across the country.

Limitations

The main limitation of this report is that the IFEM benchmarking measures were intended to provide a description of the state of a particular ED to facilitate indirect comparisons between different settings in quality and research reports, rather than being a definitive template for benchmarking quality in the New Zealand setting. The IFEM measures are only high-level and so do not drill down into details like the number of junior and senior doctors and particular models of care. This may be why individual measures did not relate strongly to the time and disposition outcomes. The total number of hospital beds may not have reflected the number of beds available for acute admissions, as there is much daily, and sometimes hourly, variation within hospitals with regard to the elective workload competing for beds and whether beds are staffed or ‘open’.

It was not possible to check the reliability of the survey data at their source, so it is possible that there were errors in the data that were provided. Some sites were not able to provide all data fields, highlighting a lack of capacity for analysis and reporting of simple metrics in some DHBs. This highlights an opportunity for investment in systems analytics in those DHBs. As this was an observational study, causality cannot be attributed to the observed associations. The small number of hospitals in New Zealand limits statistical power and the ability for more in-depth analysis.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may limit the ability of some hospitals to cope with surges in demand for acute care and impact negatively on ED performance.

Supplementary material

Summary

Abstract

Aim

The resources and capacity of New Zealand’s emergency departments (EDs) to cope with surges in demand are unknown. The aims were to describe the current resources and capacity of New Zealand EDs and explore how these relate to ED performance.

Method

A survey of EDs in New Zealand was conducted to capture elements of governance, staffing and structure of the EDs in the calendar year 2018. These were linked to processes and outcomes of care.

Results

Eighteen of 25 EDs responded. These were representative of the range of EDs nationally. There was wide variability between the EDs across all the surveyed elements. Although no single element was strongly related to performance measures, combinations of elements were. When there was a lack of doctors and available ED or hospital beds relative to the workload, then performance was worse. The correlations were: for time to assessment r=0.728, p=0.001, for ED length of stay r=0.759, p<0.001, for patients who did not wait r=0.619, p=0.006 and for deaths in the ED r=0.649, p=0.004.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may be impacting negatively on ED performance and limit the ability of some hospitals to cope with surges in demand for acute care.

Author Information

Peter Jones: Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland. Sophia Faure: Senior Advisor, Ministry of Health, Wellington. Andrew Munro: Emergency Physician, Nelson Marlborough District Health Board, Nelson.

Acknowledgements

Correspondence

A/Prof Peter Jones, Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland

Correspondence Email

Peterj@adhb.govt.nz

Competing Interests

Dr Jones reports being the Ministry of Health’s Shorter Stays in Emergency Departments health target champion. Sophia Faure reports being a Senior Advisor in the Ministry of Health.

1) Iacobucci G. Covid-19: emergency departments lack proper isolation facilities, senior medic warns. BMJ 2020;368:m953. doi: 10.1136/bmj.m953

2) Mareiniss DP. The impending storm: COVID-19, pandemics and our overwhelmed emergency departments. Am J Emerg Med 2020;38(6):1293-94. doi: 10.1016/j.ajem.2020.03.033 [published Online First: 2020/03/23]

3) Jones PG, van der Werf B. Emergency department crowding and mortality for patients presenting to emergency departments in New Zealand. Emerg Med Australas. 2021;33(4):655-64.

4) Dorsett M. Point of no return: COVID-19 and the U.S. healthcare system: An emergency physician’s perspective. Science Advances 2020;6(26):eabc5354. doi: 10.1126/sciadv.abc5354

5) Hruska K, Castrén M, Banerjee J, et al. Template for uniform reporting of emergency department measures, consensus according to the Utstein method. European Journal of Emergency Medicine 2019;26(6):417-22. doi: 10.1097/mej.0000000000000582

6) Jones PG, Mountain D, Forero R. Review article: Emergency department crowding measures associations with quality of care: A systematic review. Emerg Med Australas. 2021;epublished ahead of print, https://doi.org/10.1111/1742-6723.13743 accessed 23/03/2021.

7) Ministry of Health Manatū Hauora [Internet]. [cited 2020 Aug 06]. Available from: https://www.health.govt.nz/new-zealand-health-system/health-targets/about-health-targets/health-targets-shorter-stays-emergency-departments

8) OECD [Internet]. Hospital Beds (per 1000 inhabitants, 2019 or latest available). Available from: https://data.oecd.org/healtheqt/hospital-beds.htm accessed 23/07/2020.

9) Jones P, Wells S, Harper A, et al. Impact of a national time target for ED length of stay on patient outcomes. The New Zealand medical journal 2017;130(1455):15-34. [published Online First: 2017/05/12]

10) Tenbensel T, Chalmers L, Jones P, et al. New Zealand’s emergency department target – did it reduce ED length of stay, and if so, how and when? BMC Health Services Research 2017;17(1):678. doi: 10.1186/s12913-017-2617-1

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

In the context of the current severe acute respiratory syndrome coronavirus 2019 (COVID-19) pandemic, lack of capacity and resources for acute healthcare has been recognised internationally.[[1,2]] Many acute healthcare systems operate near, at or over capacity, with emergency department (ED) crowding and access block (admission delays for acute inpatients) a widely recognised problem worldwide. ED crowding driven by access block impairs quality of care and is a major cause of morbidity and mortality for acute patients.[[3]] It is unclear whether New Zealand’s EDs are prepared for a ‘slow moving mass casualty incident’, such as has been seen in countries like the USA over the past 18 months.[[4]] Another unknown is whether EDs in New Zealand have comparable staffing and capacity, so some regions may be more able to cope with such disasters than others. To facilitate understanding and interpretation of international comparisons of acute care, the International Federation for Emergency Medicine (IFEM) recently developed a template for uniform reporting of ED structure, staffing, governance and performance.[[5]] This survey includes the performance measures time to assessment, which is related to patient satisfaction and experience, and ED length of stay (LOS), which is related to patient outcomes including mortality.[[6,7]]

The primary aim of this study was to document the current state of EDs in New Zealand according to the IFEM template and to look for opportunities to reduce variation nationally. The secondary aim was to explore whether any of the structural and staffing variables were associated with the process and outcome measures.

Methods

This was a retrospective cross-sectional survey of EDs in New Zealand for the calendar year 2018.

Site selection

All 25 EDs in New Zealand that report data to the Ministry of Health (MOH) for the SSED target were eligible to participate. Participants were invited to contribute data via email to the ED clinical director. Participation was voluntary and not incentivised.

Survey instrument

The survey questions were those suggested by the IFEM.[[5]] The template has sections based on ED structure, staffing and governance, attending population, processes and outcomes. In acknowledgement that ED processes and outcomes depend on the bed-state of the hospital, the template also includes a section on the number of hospital beds for the population served by the hospital.[[5]]

Data collection and checking

An electronic link to an online survey managed by the MOH was provided, and on request a Microsoft Word™ document was also available for manual data entry. A manual with data definitions was provided with the survey instruments. The survey opened for responses on 22 May 2019 and closed on 22 July 2019. Survey questions are shown in the supplementary material. Where data were incomplete or incongruent, follow-up emails were sent with a request to complete or check the validity of the data. Data on hospital bed numbers were also cross checked with individual district health board (DHB) websites and the MOH website for each DHB. Where individual hospital populations could not be determined, the DHB population and total DHB hospital bed numbers were used in the calculation of beds/1,000 population. The correlation between the bed numbers reported from survey respondents and the MOH website was r=0.99, with the MOH website estimating 5.2% more beds than the survey respondents.

Data analysis

Data were downloaded and entered into a spreadsheet (Microsoft Excel™ 2016). Analysis was descriptive, using proportion with 95% confidence interval (95%CI), median with interquartile range (IQR) or mean with standard deviation.

The relationship between variables was assessed visually using scatter plots. Where appropriate, correlations were performed using SPSSv22 (IBM corporation, Armonk, NY, USA). The distribution of each variable was assessed visually using histograms and Q-Q plots and statistically with the Kolmogorov–Smirnoff test. Pearson correlation was used for normally distributed variables; otherwise Spearman’s correlation was used. Statistical significance was adjusted for multiple comparisons using the Bonferroni calculation (alpha/number of comparisons), so statistical significance was set at p<0.001. A sample size calculation was not performed as this was an exploratory survey with a maximum sample size of 25.

Ethics

As a survey of administrative data, this study was out of scope for the New Zealand Health and Disability Ethics Committees.

Results

Responses were received from 18/26 hospitals (69%), which together were representative of smaller regional hospitals through to major urban tertiary hospitals. The number of hospital beds ranged from 43 to 1,005, and hospital beds per 1,000 population ranged from 1.5 to 2.3. The number of presentations ranged from fewer than 20,000 per year to more than 100,000, with the presentations per 1,000 DHB population ranging from 163 to 352 (Table 2).

Governance, staffing and structure

Table 1 shows the variation in size, structure and staffing of EDs. The number of beds ranged from 12 with no ED short stay unit to 57 with a 23-bed short stay unit. Staffing also varied: 105 to 259 nursing hours per 100 patient visits; 0 to 28 advanced care (nurse) practitioner hours per 100 patient visits; and 68.7 to 146 physician hours per 100 patient visits. Only one ED reported direct admitting rights, and one reported having an emergency medicine specialist in the ED 24/7.

Table 1: Structure, staffing and governance. View Table 1.

Population presenting to ED

Table 2 shows the variation in patient age and acuity of ED presentations to the different departments. Excluding Starship Children’s Hospital and Auckland City Hospital, which see only children or adults respectively, the proportion of children aged 0–5 ranged from 4.2% to 19.4%. The number of adults over 75 years ranged from 9.6% to 19.3%. There was also a wide variation in the proportion of ambulance arrivals (7.2% to 33.3%). Triage acuity also varied. One marked difference was in the higher proportion of resuscitation cases that presented to Auckland City Hospital (more than 4%) when all other departments saw <1% (Table 2).

Table 2: Presenting population. View Table 2.

Processes and outcomes

Table 3 shows ED process times. The wait to be seen ranged from median of 13 (IQR 6, 37) to 92 (IQR 41, 162) minutes. The ED LOS varied from a median of 135 (IQR 76, 222) to 288 (IQR 177, 517) minutes. With respect to outcomes, patients not waiting to be seen ranged from 1.0% to 8.0%, whereas admission rates ranged from 9.6 to 40.0%. There were few deaths in the ED (<0.01% to 0.14%). The proportion of patients who re-presented to the ED within 72 hours and were subsequently admitted ranged from 0.6% to 5.1%, and all but two sites reported proportions less than 2%.

Table 3: Processes and outcomes. View Table 3.

Relationship of structure, workload and staffing to processes and outcomes

Structural measures

Examination of the scatterplots showed no linear relationship between either ED spaces per 1,000 presentations or hospital beds per 1,000 DHB population for any process or outcome (see the supplementary material).

Workload measures

The scatterplots suggested a there may be a relationship between the proportion of ambulance presentations and most processes and outcomes. There were some moderate correlations:

  • time to assessment r=0.590, p=0.010
  • ED LOS r=0.593, p=0.012
  • did not wait (DNW) r=0.566, p=014
  • deaths in ED r=0.513, p=0.029

Although the proportion of admissions was uncorrelated to outcomes (scatterplots shown in the supplementary material), the proportion of admissions was moderately correlated to ED processes:

  • time to assessment r=0.494, p=0.037
  • ED LOS r=0.609, p=0.009

The proportion of high acuity presentations was associated with the proportion who died in ED (r=0.564, p=0.015) but no other processes or outcomes.

Staff measures

The scatterplots showed no linear relationship between nursing hours or advanced care practitioners per 100 patients and any of the process or outcome measures (see the supplementary material). Doctor hours per 100 patients was negatively correlated to time to assessment (r=-0.617, p=0.006) and DNW (r=-0.515, p=0.029), although not to other processes and outcomes (scatterplots shown in the supplementary material).

Combinations of measures

Exploratory analysis looking at different combinations of structural, workload and staffing measures found stronger correlations with process and outcome measures than with the individual variables:

  • For time to assessment, the strongest correlation was seen when sites had high ambulance presentations, high admissions to hospital and low doctor hours: r=0.728, p=0.001.
  • For ED LOS, the strongest correlation was with high ambulance presentations, high admission rates, fewer beds per 1,000 DHB population and low doctor hours: r=0.759, p<0.001.
  • For DNW, the strongest correlation was with high ambulance presentations and fewer ED treatment spaces per 100,000 DHB population: r=0.619, p=0.006.
  • For deaths in the ED, the strongest correlation was for high ambulance presentations, fewer ED treatment spaces per 100,000 DHB population and low doctor hours: r=0.649, p=0.004.

Figure 1 shows the scatterplots for these associations and the full table of all correlations tested is provided in the supplementary material.

The scatterplots showed that the proportion of patients re-presenting and being re-admitted within 72 hours was not linearly related to any structure, workload or staffing measure, or any combination of these measures, as most sites had fewer than 2% re-presentations requiring admission, with two notable outliers heavily influencing the observed correlations.

Discussion

The survey found marked variation in the presenting population, structure, staffing and governance of New Zealand EDs. The reasons for the variation are likely to be multifactorial and may be inter-related. Some of the variation is most likely due to the locations of hospitals and the availability of alternative care. For example, the populations presenting to smaller rural or regional hospitals that have limited availability of after-hours primary or urgent care was more likely to be lower acuity, since the only available option for care is the ED. This was different in major centres, where urgent care clinics deal with the vast majority of minor injury and medical complaints. This was reflected in the lower presentation rates, higher admission to hospital rates and corresponding longer lengths of stay observed in larger urban tertiary hospitals compared to rural and regional centres. A notable exception among tertiary urban referral hospitals was the sole specialist paediatric hospital, which had fewer patients, lower triage acuity, higher staffing, shorter times to assessment, fewer admissions and few patients leaving prior to assessment. In contrast, the sole adult hospital had a high proportion of triage 1 cases, which reflected its role as the regional trauma, stroke and 24-hour interventional cardiology centre.

There was marked variation in the number of bed spaces per 1,000 presentations, ranging from under 0.6 to more than 1.0. Sites with fewer ED beds per 1,000 presentations were likely to be more fully occupied more of the time and be less able to cope with surges in demand, such as has been seen internationally during the COVID-19 pandemic. Similarly, hospital bed numbers per 1,000 population varied considerably. The average of 1.9 was significantly lower than 2.6 of beds per 1,000 population reported for New Zealand by the Organisation for Economic Cooperation and Development (OECD).[[8]] In the context of a national disaster or pandemic, those hospitals with lower bed number per head of population will less likely cope with surges in demand. In sites with few ED spaces and very low hospital bed numbers per 1,000, there is an opportunity for infrastructure investment to redress this deficiency.

There was wide variation in nursing hours per 100 presentations. No clear pattern with respect to number of presentations was observed. Physician plus advanced care provider ratios per 100 patients varied less, except for the two busiest EDs, which had had the lowest clinician-to-patient ratios. A notable exception was the specialist paediatric hospital, which had much higher staffing ratios than other sites.

Most EDs had a short stay unit run by an emergency medicine specialist, which allowed for longer periods of ED care and prevented unnecessary hospital admissions. The use of ED short stay units increased considerably in the last decade, partially in response to the SSED health target.[[9]]

The reporting measures used in isolation were not associated strongly with the observed variation in performance. Only the proportion of ambulance presentations seemed possibly related to most ED process and outcome measures. The proportion of admissions was also related to ED processes, although the observed relationships were not strong. Unsurprisingly, a higher proportion of ambulance arrivals was associated with more deaths in the ED: patients who arrive by ambulance are typically more unwell. The impact of ambulance arrivals on ED process measures may have reflected the extra time required to investigate and treat these especially unwell patients. It was also recognised that the efficiency of admission processes within the hospital has an influence on ED LOS for admitted patients.[[10]]

When structural, staffing and workload measures were combined, there were clearer relationships between the processes and outcomes of care. When sites with higher proportions of ambulance arrivals and higher admission rates also had fewer doctors, less space or fewer hospital beds, then there were prolonged times to assessment, longer ED LOS, more patients leaving without being seen and more deaths in the ED. Investment in infrastructure and staffing may be required at these sites to improve processes and outcomes of care and reduce variation across the country.

Limitations

The main limitation of this report is that the IFEM benchmarking measures were intended to provide a description of the state of a particular ED to facilitate indirect comparisons between different settings in quality and research reports, rather than being a definitive template for benchmarking quality in the New Zealand setting. The IFEM measures are only high-level and so do not drill down into details like the number of junior and senior doctors and particular models of care. This may be why individual measures did not relate strongly to the time and disposition outcomes. The total number of hospital beds may not have reflected the number of beds available for acute admissions, as there is much daily, and sometimes hourly, variation within hospitals with regard to the elective workload competing for beds and whether beds are staffed or ‘open’.

It was not possible to check the reliability of the survey data at their source, so it is possible that there were errors in the data that were provided. Some sites were not able to provide all data fields, highlighting a lack of capacity for analysis and reporting of simple metrics in some DHBs. This highlights an opportunity for investment in systems analytics in those DHBs. As this was an observational study, causality cannot be attributed to the observed associations. The small number of hospitals in New Zealand limits statistical power and the ability for more in-depth analysis.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may limit the ability of some hospitals to cope with surges in demand for acute care and impact negatively on ED performance.

Supplementary material

Summary

Abstract

Aim

The resources and capacity of New Zealand’s emergency departments (EDs) to cope with surges in demand are unknown. The aims were to describe the current resources and capacity of New Zealand EDs and explore how these relate to ED performance.

Method

A survey of EDs in New Zealand was conducted to capture elements of governance, staffing and structure of the EDs in the calendar year 2018. These were linked to processes and outcomes of care.

Results

Eighteen of 25 EDs responded. These were representative of the range of EDs nationally. There was wide variability between the EDs across all the surveyed elements. Although no single element was strongly related to performance measures, combinations of elements were. When there was a lack of doctors and available ED or hospital beds relative to the workload, then performance was worse. The correlations were: for time to assessment r=0.728, p=0.001, for ED length of stay r=0.759, p<0.001, for patients who did not wait r=0.619, p=0.006 and for deaths in the ED r=0.649, p=0.004.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may be impacting negatively on ED performance and limit the ability of some hospitals to cope with surges in demand for acute care.

Author Information

Peter Jones: Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland. Sophia Faure: Senior Advisor, Ministry of Health, Wellington. Andrew Munro: Emergency Physician, Nelson Marlborough District Health Board, Nelson.

Acknowledgements

Correspondence

A/Prof Peter Jones, Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland

Correspondence Email

Peterj@adhb.govt.nz

Competing Interests

Dr Jones reports being the Ministry of Health’s Shorter Stays in Emergency Departments health target champion. Sophia Faure reports being a Senior Advisor in the Ministry of Health.

1) Iacobucci G. Covid-19: emergency departments lack proper isolation facilities, senior medic warns. BMJ 2020;368:m953. doi: 10.1136/bmj.m953

2) Mareiniss DP. The impending storm: COVID-19, pandemics and our overwhelmed emergency departments. Am J Emerg Med 2020;38(6):1293-94. doi: 10.1016/j.ajem.2020.03.033 [published Online First: 2020/03/23]

3) Jones PG, van der Werf B. Emergency department crowding and mortality for patients presenting to emergency departments in New Zealand. Emerg Med Australas. 2021;33(4):655-64.

4) Dorsett M. Point of no return: COVID-19 and the U.S. healthcare system: An emergency physician’s perspective. Science Advances 2020;6(26):eabc5354. doi: 10.1126/sciadv.abc5354

5) Hruska K, Castrén M, Banerjee J, et al. Template for uniform reporting of emergency department measures, consensus according to the Utstein method. European Journal of Emergency Medicine 2019;26(6):417-22. doi: 10.1097/mej.0000000000000582

6) Jones PG, Mountain D, Forero R. Review article: Emergency department crowding measures associations with quality of care: A systematic review. Emerg Med Australas. 2021;epublished ahead of print, https://doi.org/10.1111/1742-6723.13743 accessed 23/03/2021.

7) Ministry of Health Manatū Hauora [Internet]. [cited 2020 Aug 06]. Available from: https://www.health.govt.nz/new-zealand-health-system/health-targets/about-health-targets/health-targets-shorter-stays-emergency-departments

8) OECD [Internet]. Hospital Beds (per 1000 inhabitants, 2019 or latest available). Available from: https://data.oecd.org/healtheqt/hospital-beds.htm accessed 23/07/2020.

9) Jones P, Wells S, Harper A, et al. Impact of a national time target for ED length of stay on patient outcomes. The New Zealand medical journal 2017;130(1455):15-34. [published Online First: 2017/05/12]

10) Tenbensel T, Chalmers L, Jones P, et al. New Zealand’s emergency department target – did it reduce ED length of stay, and if so, how and when? BMC Health Services Research 2017;17(1):678. doi: 10.1186/s12913-017-2617-1

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

In the context of the current severe acute respiratory syndrome coronavirus 2019 (COVID-19) pandemic, lack of capacity and resources for acute healthcare has been recognised internationally.[[1,2]] Many acute healthcare systems operate near, at or over capacity, with emergency department (ED) crowding and access block (admission delays for acute inpatients) a widely recognised problem worldwide. ED crowding driven by access block impairs quality of care and is a major cause of morbidity and mortality for acute patients.[[3]] It is unclear whether New Zealand’s EDs are prepared for a ‘slow moving mass casualty incident’, such as has been seen in countries like the USA over the past 18 months.[[4]] Another unknown is whether EDs in New Zealand have comparable staffing and capacity, so some regions may be more able to cope with such disasters than others. To facilitate understanding and interpretation of international comparisons of acute care, the International Federation for Emergency Medicine (IFEM) recently developed a template for uniform reporting of ED structure, staffing, governance and performance.[[5]] This survey includes the performance measures time to assessment, which is related to patient satisfaction and experience, and ED length of stay (LOS), which is related to patient outcomes including mortality.[[6,7]]

The primary aim of this study was to document the current state of EDs in New Zealand according to the IFEM template and to look for opportunities to reduce variation nationally. The secondary aim was to explore whether any of the structural and staffing variables were associated with the process and outcome measures.

Methods

This was a retrospective cross-sectional survey of EDs in New Zealand for the calendar year 2018.

Site selection

All 25 EDs in New Zealand that report data to the Ministry of Health (MOH) for the SSED target were eligible to participate. Participants were invited to contribute data via email to the ED clinical director. Participation was voluntary and not incentivised.

Survey instrument

The survey questions were those suggested by the IFEM.[[5]] The template has sections based on ED structure, staffing and governance, attending population, processes and outcomes. In acknowledgement that ED processes and outcomes depend on the bed-state of the hospital, the template also includes a section on the number of hospital beds for the population served by the hospital.[[5]]

Data collection and checking

An electronic link to an online survey managed by the MOH was provided, and on request a Microsoft Word™ document was also available for manual data entry. A manual with data definitions was provided with the survey instruments. The survey opened for responses on 22 May 2019 and closed on 22 July 2019. Survey questions are shown in the supplementary material. Where data were incomplete or incongruent, follow-up emails were sent with a request to complete or check the validity of the data. Data on hospital bed numbers were also cross checked with individual district health board (DHB) websites and the MOH website for each DHB. Where individual hospital populations could not be determined, the DHB population and total DHB hospital bed numbers were used in the calculation of beds/1,000 population. The correlation between the bed numbers reported from survey respondents and the MOH website was r=0.99, with the MOH website estimating 5.2% more beds than the survey respondents.

Data analysis

Data were downloaded and entered into a spreadsheet (Microsoft Excel™ 2016). Analysis was descriptive, using proportion with 95% confidence interval (95%CI), median with interquartile range (IQR) or mean with standard deviation.

The relationship between variables was assessed visually using scatter plots. Where appropriate, correlations were performed using SPSSv22 (IBM corporation, Armonk, NY, USA). The distribution of each variable was assessed visually using histograms and Q-Q plots and statistically with the Kolmogorov–Smirnoff test. Pearson correlation was used for normally distributed variables; otherwise Spearman’s correlation was used. Statistical significance was adjusted for multiple comparisons using the Bonferroni calculation (alpha/number of comparisons), so statistical significance was set at p<0.001. A sample size calculation was not performed as this was an exploratory survey with a maximum sample size of 25.

Ethics

As a survey of administrative data, this study was out of scope for the New Zealand Health and Disability Ethics Committees.

Results

Responses were received from 18/26 hospitals (69%), which together were representative of smaller regional hospitals through to major urban tertiary hospitals. The number of hospital beds ranged from 43 to 1,005, and hospital beds per 1,000 population ranged from 1.5 to 2.3. The number of presentations ranged from fewer than 20,000 per year to more than 100,000, with the presentations per 1,000 DHB population ranging from 163 to 352 (Table 2).

Governance, staffing and structure

Table 1 shows the variation in size, structure and staffing of EDs. The number of beds ranged from 12 with no ED short stay unit to 57 with a 23-bed short stay unit. Staffing also varied: 105 to 259 nursing hours per 100 patient visits; 0 to 28 advanced care (nurse) practitioner hours per 100 patient visits; and 68.7 to 146 physician hours per 100 patient visits. Only one ED reported direct admitting rights, and one reported having an emergency medicine specialist in the ED 24/7.

Table 1: Structure, staffing and governance. View Table 1.

Population presenting to ED

Table 2 shows the variation in patient age and acuity of ED presentations to the different departments. Excluding Starship Children’s Hospital and Auckland City Hospital, which see only children or adults respectively, the proportion of children aged 0–5 ranged from 4.2% to 19.4%. The number of adults over 75 years ranged from 9.6% to 19.3%. There was also a wide variation in the proportion of ambulance arrivals (7.2% to 33.3%). Triage acuity also varied. One marked difference was in the higher proportion of resuscitation cases that presented to Auckland City Hospital (more than 4%) when all other departments saw <1% (Table 2).

Table 2: Presenting population. View Table 2.

Processes and outcomes

Table 3 shows ED process times. The wait to be seen ranged from median of 13 (IQR 6, 37) to 92 (IQR 41, 162) minutes. The ED LOS varied from a median of 135 (IQR 76, 222) to 288 (IQR 177, 517) minutes. With respect to outcomes, patients not waiting to be seen ranged from 1.0% to 8.0%, whereas admission rates ranged from 9.6 to 40.0%. There were few deaths in the ED (<0.01% to 0.14%). The proportion of patients who re-presented to the ED within 72 hours and were subsequently admitted ranged from 0.6% to 5.1%, and all but two sites reported proportions less than 2%.

Table 3: Processes and outcomes. View Table 3.

Relationship of structure, workload and staffing to processes and outcomes

Structural measures

Examination of the scatterplots showed no linear relationship between either ED spaces per 1,000 presentations or hospital beds per 1,000 DHB population for any process or outcome (see the supplementary material).

Workload measures

The scatterplots suggested a there may be a relationship between the proportion of ambulance presentations and most processes and outcomes. There were some moderate correlations:

  • time to assessment r=0.590, p=0.010
  • ED LOS r=0.593, p=0.012
  • did not wait (DNW) r=0.566, p=014
  • deaths in ED r=0.513, p=0.029

Although the proportion of admissions was uncorrelated to outcomes (scatterplots shown in the supplementary material), the proportion of admissions was moderately correlated to ED processes:

  • time to assessment r=0.494, p=0.037
  • ED LOS r=0.609, p=0.009

The proportion of high acuity presentations was associated with the proportion who died in ED (r=0.564, p=0.015) but no other processes or outcomes.

Staff measures

The scatterplots showed no linear relationship between nursing hours or advanced care practitioners per 100 patients and any of the process or outcome measures (see the supplementary material). Doctor hours per 100 patients was negatively correlated to time to assessment (r=-0.617, p=0.006) and DNW (r=-0.515, p=0.029), although not to other processes and outcomes (scatterplots shown in the supplementary material).

Combinations of measures

Exploratory analysis looking at different combinations of structural, workload and staffing measures found stronger correlations with process and outcome measures than with the individual variables:

  • For time to assessment, the strongest correlation was seen when sites had high ambulance presentations, high admissions to hospital and low doctor hours: r=0.728, p=0.001.
  • For ED LOS, the strongest correlation was with high ambulance presentations, high admission rates, fewer beds per 1,000 DHB population and low doctor hours: r=0.759, p<0.001.
  • For DNW, the strongest correlation was with high ambulance presentations and fewer ED treatment spaces per 100,000 DHB population: r=0.619, p=0.006.
  • For deaths in the ED, the strongest correlation was for high ambulance presentations, fewer ED treatment spaces per 100,000 DHB population and low doctor hours: r=0.649, p=0.004.

Figure 1 shows the scatterplots for these associations and the full table of all correlations tested is provided in the supplementary material.

The scatterplots showed that the proportion of patients re-presenting and being re-admitted within 72 hours was not linearly related to any structure, workload or staffing measure, or any combination of these measures, as most sites had fewer than 2% re-presentations requiring admission, with two notable outliers heavily influencing the observed correlations.

Discussion

The survey found marked variation in the presenting population, structure, staffing and governance of New Zealand EDs. The reasons for the variation are likely to be multifactorial and may be inter-related. Some of the variation is most likely due to the locations of hospitals and the availability of alternative care. For example, the populations presenting to smaller rural or regional hospitals that have limited availability of after-hours primary or urgent care was more likely to be lower acuity, since the only available option for care is the ED. This was different in major centres, where urgent care clinics deal with the vast majority of minor injury and medical complaints. This was reflected in the lower presentation rates, higher admission to hospital rates and corresponding longer lengths of stay observed in larger urban tertiary hospitals compared to rural and regional centres. A notable exception among tertiary urban referral hospitals was the sole specialist paediatric hospital, which had fewer patients, lower triage acuity, higher staffing, shorter times to assessment, fewer admissions and few patients leaving prior to assessment. In contrast, the sole adult hospital had a high proportion of triage 1 cases, which reflected its role as the regional trauma, stroke and 24-hour interventional cardiology centre.

There was marked variation in the number of bed spaces per 1,000 presentations, ranging from under 0.6 to more than 1.0. Sites with fewer ED beds per 1,000 presentations were likely to be more fully occupied more of the time and be less able to cope with surges in demand, such as has been seen internationally during the COVID-19 pandemic. Similarly, hospital bed numbers per 1,000 population varied considerably. The average of 1.9 was significantly lower than 2.6 of beds per 1,000 population reported for New Zealand by the Organisation for Economic Cooperation and Development (OECD).[[8]] In the context of a national disaster or pandemic, those hospitals with lower bed number per head of population will less likely cope with surges in demand. In sites with few ED spaces and very low hospital bed numbers per 1,000, there is an opportunity for infrastructure investment to redress this deficiency.

There was wide variation in nursing hours per 100 presentations. No clear pattern with respect to number of presentations was observed. Physician plus advanced care provider ratios per 100 patients varied less, except for the two busiest EDs, which had had the lowest clinician-to-patient ratios. A notable exception was the specialist paediatric hospital, which had much higher staffing ratios than other sites.

Most EDs had a short stay unit run by an emergency medicine specialist, which allowed for longer periods of ED care and prevented unnecessary hospital admissions. The use of ED short stay units increased considerably in the last decade, partially in response to the SSED health target.[[9]]

The reporting measures used in isolation were not associated strongly with the observed variation in performance. Only the proportion of ambulance presentations seemed possibly related to most ED process and outcome measures. The proportion of admissions was also related to ED processes, although the observed relationships were not strong. Unsurprisingly, a higher proportion of ambulance arrivals was associated with more deaths in the ED: patients who arrive by ambulance are typically more unwell. The impact of ambulance arrivals on ED process measures may have reflected the extra time required to investigate and treat these especially unwell patients. It was also recognised that the efficiency of admission processes within the hospital has an influence on ED LOS for admitted patients.[[10]]

When structural, staffing and workload measures were combined, there were clearer relationships between the processes and outcomes of care. When sites with higher proportions of ambulance arrivals and higher admission rates also had fewer doctors, less space or fewer hospital beds, then there were prolonged times to assessment, longer ED LOS, more patients leaving without being seen and more deaths in the ED. Investment in infrastructure and staffing may be required at these sites to improve processes and outcomes of care and reduce variation across the country.

Limitations

The main limitation of this report is that the IFEM benchmarking measures were intended to provide a description of the state of a particular ED to facilitate indirect comparisons between different settings in quality and research reports, rather than being a definitive template for benchmarking quality in the New Zealand setting. The IFEM measures are only high-level and so do not drill down into details like the number of junior and senior doctors and particular models of care. This may be why individual measures did not relate strongly to the time and disposition outcomes. The total number of hospital beds may not have reflected the number of beds available for acute admissions, as there is much daily, and sometimes hourly, variation within hospitals with regard to the elective workload competing for beds and whether beds are staffed or ‘open’.

It was not possible to check the reliability of the survey data at their source, so it is possible that there were errors in the data that were provided. Some sites were not able to provide all data fields, highlighting a lack of capacity for analysis and reporting of simple metrics in some DHBs. This highlights an opportunity for investment in systems analytics in those DHBs. As this was an observational study, causality cannot be attributed to the observed associations. The small number of hospitals in New Zealand limits statistical power and the ability for more in-depth analysis.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may limit the ability of some hospitals to cope with surges in demand for acute care and impact negatively on ED performance.

Supplementary material

Summary

Abstract

Aim

The resources and capacity of New Zealand’s emergency departments (EDs) to cope with surges in demand are unknown. The aims were to describe the current resources and capacity of New Zealand EDs and explore how these relate to ED performance.

Method

A survey of EDs in New Zealand was conducted to capture elements of governance, staffing and structure of the EDs in the calendar year 2018. These were linked to processes and outcomes of care.

Results

Eighteen of 25 EDs responded. These were representative of the range of EDs nationally. There was wide variability between the EDs across all the surveyed elements. Although no single element was strongly related to performance measures, combinations of elements were. When there was a lack of doctors and available ED or hospital beds relative to the workload, then performance was worse. The correlations were: for time to assessment r=0.728, p=0.001, for ED length of stay r=0.759, p<0.001, for patients who did not wait r=0.619, p=0.006 and for deaths in the ED r=0.649, p=0.004.

Conclusion

There is marked variation among New Zealand hospitals with respect to structure, staffing and workload, which may be impacting negatively on ED performance and limit the ability of some hospitals to cope with surges in demand for acute care.

Author Information

Peter Jones: Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland. Sophia Faure: Senior Advisor, Ministry of Health, Wellington. Andrew Munro: Emergency Physician, Nelson Marlborough District Health Board, Nelson.

Acknowledgements

Correspondence

A/Prof Peter Jones, Emergency Physician, Adult Emergency Department, Auckland City Hospital, Auckland District Health Board, Department of Surgery, School of Medicine, University of Auckland

Correspondence Email

Peterj@adhb.govt.nz

Competing Interests

Dr Jones reports being the Ministry of Health’s Shorter Stays in Emergency Departments health target champion. Sophia Faure reports being a Senior Advisor in the Ministry of Health.

1) Iacobucci G. Covid-19: emergency departments lack proper isolation facilities, senior medic warns. BMJ 2020;368:m953. doi: 10.1136/bmj.m953

2) Mareiniss DP. The impending storm: COVID-19, pandemics and our overwhelmed emergency departments. Am J Emerg Med 2020;38(6):1293-94. doi: 10.1016/j.ajem.2020.03.033 [published Online First: 2020/03/23]

3) Jones PG, van der Werf B. Emergency department crowding and mortality for patients presenting to emergency departments in New Zealand. Emerg Med Australas. 2021;33(4):655-64.

4) Dorsett M. Point of no return: COVID-19 and the U.S. healthcare system: An emergency physician’s perspective. Science Advances 2020;6(26):eabc5354. doi: 10.1126/sciadv.abc5354

5) Hruska K, Castrén M, Banerjee J, et al. Template for uniform reporting of emergency department measures, consensus according to the Utstein method. European Journal of Emergency Medicine 2019;26(6):417-22. doi: 10.1097/mej.0000000000000582

6) Jones PG, Mountain D, Forero R. Review article: Emergency department crowding measures associations with quality of care: A systematic review. Emerg Med Australas. 2021;epublished ahead of print, https://doi.org/10.1111/1742-6723.13743 accessed 23/03/2021.

7) Ministry of Health Manatū Hauora [Internet]. [cited 2020 Aug 06]. Available from: https://www.health.govt.nz/new-zealand-health-system/health-targets/about-health-targets/health-targets-shorter-stays-emergency-departments

8) OECD [Internet]. Hospital Beds (per 1000 inhabitants, 2019 or latest available). Available from: https://data.oecd.org/healtheqt/hospital-beds.htm accessed 23/07/2020.

9) Jones P, Wells S, Harper A, et al. Impact of a national time target for ED length of stay on patient outcomes. The New Zealand medical journal 2017;130(1455):15-34. [published Online First: 2017/05/12]

10) Tenbensel T, Chalmers L, Jones P, et al. New Zealand’s emergency department target – did it reduce ED length of stay, and if so, how and when? BMC Health Services Research 2017;17(1):678. doi: 10.1186/s12913-017-2617-1

Contact diana@nzma.org.nz
for the PDF of this article

Subscriber Content

The full contents of this pages only available to subscribers.
Login, subscribe or email nzmj@nzma.org.nz to purchase this article.

LOGINSUBSCRIBE