Journal of the New Zealand Medical Association, 26-October-2007, Vol 120 No 1264
Trends in hospital bed utilisation in New Zealand 1989 to 2006: more or less beds in the future?
Hospital beds are an important and costly resource for all health systems. Attempts to contain health expenditure for the past 2–3 decades have focused strongly upon reducing both bed numbers and utilisation through reducing length of stay, admission gate-keeping processes, and the provision of alternative community-based services1. As a consequence, there has been a progressive reduction in bed rate availability and utilisation in almost all health systems.
District health boards (DHBs), with their integrated hospital/secondary and primary care/community-based services, provide an opportunity open to few other health systems to develop alternatives to expensive hospital-based care. With increasing pressures upon DHB resources and the emergence of an organized primary health care system, several initiatives have been established through DHBs in conjunction with independent practice associations (IPAs) and more recently primary health organisations (PHOs) to develop more substantial community-based alternatives to hospital care.
In New Zealand, downward trends in bed availability and bed days have been apparent for at least two decades in hospital boards and more recently in DHBs. This has been promoted by population-based funding.2 Although the Ministry of Health for nearly two decades has published data on hospital utilisation/throughput, there has been very little analysis of the trends from these data.3
This paper therefore has the following aims: to analyse trends in hospital bed utilisation for bed availability, bed days, discharge rates, and trends in day patients from 1988 to 2006; to examine variation between DHBs in utilisation; to consider factors which may explain the trends observed; to compare the results with international trends in bed utilisation (particularly Australia, England, and the US), and to discuss future need for beds and factors affecting this need, including the integrated DHB system.
The following sources of data and information were used to prepare this paper.
Ministry of Health Hospital Throughput reports were obtained from 1988/89 to 2005/06.3 Full details of how the data are prepared for comparison are presented in the Appendices to these reports. For the most part, these reports are on the Ministry of Health website but for the 3 years ending June 2006 were obtained directly from Ministry staff. These reports present data on hospital utilisation in detail and trend data. The data were used to analyse trends in discharges/1000 population, day patients, bed days, and average length of stay (ALOS) and bed availability. Allowance for bed days and numbers was made for day patients by allocating one bed day per day patient.
Ministry of Health data were sought for numbers of licensed beds, both public and private. However the latest available figures are only for 2001. Hence current bed availability was calculated assuming 85% occupancy. New Zealand Private Surgical Hospitals Association provided private hospital data from reports from its membership for the 2006/07. Data for some previous years was derived from reports and surveys of the Private Hospitals Association.
New Zealand Hospital Throughput data have been case weighted, ‘filtered’, and ‘truncated’ to ensure as far as possible comparability between DHBs and their predecessors and trends over time. Filtering removes certain DRG categories such as well babies, short-stay observations, endoscopies, some mental health events, and other categories totalling about 26. Truncation removes bed days above the 97th percentile for each DRG. The overall effect is to reduce discharge numbers by about 3% and bed days by 6%.
Comparison between years and a DHBs were based upon standardised data.3 Two methods were used by the MOH. For comparison by DHB regions, age/socioeconomic deprivation/ethnic standardised nation was employed. For national comparisons over time, and comparisons between socioeconomic deprivation groups, age/gender standardisation was used.3
Casemix adjustment takes into account changes in the types of treatment that patients receive which vary in complexity and cost. Comparisons between DHBs and over time were based upon national hospital reference prices.3 This takes into account the change in ICD coding from July 2000 and other changes, e.g. in data quality.
International sources included data from the Australian Institute of Health and Welfare4 and a comparison of hospital utilisation in the UK NHS and the Californian-based HMO Kaiser Permanente by Feachem et al.5
Table 1 presents calculated results of rates of bed availability and utilisation annually for the period 1988/89 to 2005/06. Trends in these indicators are also presented in Figures 1–4. The following findings should be noted from Table 1 and the Figures.
Discharge rates—Total discharge rates, and especially standardised discharge rates, have steady increased until 2000/01 but since then the rate has flattened off. However this increase is almost entirely due to the rate of increase in day patients with no increase over this period in inpatient rates. The larger increase in standardised rates indicates an increase in acuity of patients receiving hospital care.
Average length of stay (ALOS)—Figure 2 shows at a continuing steady decline over the period in both inpatient and total ALOS. In 1988/89, the inpatient ALOS was 6.67 and the total 6.13 days. In 2005/2006 these figures had declined to 3.90 and 2.81 days respectively i.e. a more than 50% drop in the total figure.
Rate of bed days utilisation—Figure 3 shows an associated decline in inpatient bed days. As described above these are truncated for the purposes of comparison between years. Table 1 also shows an increasing number of the day patients in which one-bed day was allocated for each day patient.
Rate of bed availability—Figure 4 shows the calculated availability of beds in both public and private sectors. As indicated above data are not available from the Ministry of Health for public bed availability since 2000. In any case the figures provided in the past by the MOH have been based upon licensed beds and have not excluded beds used for longer stay patients, including mental patients. Hence they cannot be appropriately compared for trends with the truncated and filtered Hospital Throughput data. The trends of Table 1 and Figure 4 show a decline in public bed availability to a current 1.56/1000 with some increase in the private sector to 0.4 giving a total of nearly 2.0/1000 population.
Variation between DHBs—Figure 5 shows the variation between DHBs in discharge ratios standardised for age, gender, and casemix for the 2005/2006 year. There is obviously a wide variation between DHBs from a low of 0.84 in Capital and Coast DHB to a high of 1.20 in Tairawhiti and Wairarapa DHBs—i.e. a ratio of 43%. In other words even after allowing for need factors some DHBs have very much lower and higher discharge rates than others.
Ministry of Health calculations indicate that variations of more than about 4% from the mean of 1.00 are statistically highly significant at 99% confidence limits. Examination of the equivalent data for 2001/2002 shows that this variation has markedly increased.
Table 2 compares the most recent New Zealand data Australian sources for 2005 (Australian Institute of Health and Welfare)4 and data from a comparative study of the UK NHS with the California-based HMO Kaiser Permanente for the year 2000.5,6 Although not specifically stated it appears that this comparison includes only inpatients and excludes day patients.
Table 2. Comparison of rates/1000 population for bed utilisation in New Zealand (2005/06), Australia (2005), NHS (2000), and US Kaiser Permanente (2000)
Given the different years of comparison and trends over previous years it is likely that both the NHS and Kaiser figures for ALOS and acute bed days and probably bed availability are likely to be even less than those presented.
Although there are uncertainties in these comparisons, with doubts about what is included and excluded, they clearly indicate that New Zealand has a much lower rate of hospital utilisation than Australia and the NHS but a much higher rate than Kaiser Permanente. The latter figures clearly indicate that the much lower rate of bed utilization, less than 50%, of the figures for New Zealand is due to much lower discharge rates than the ALOS.
Figure 1. Trends in rates of throughput (discharges) per 1000 population of public and total daypatients and inpatients, inclusive of daypatients from 1998/89 to 2005/06
Figure 2. Trends in average length of stay of inpatients (truncated) and total patients including day patients from 1998/89 to 2005/06
Figure 3. Trends in rates for New Zealand per 1000 population of public bed days, inpatient, and day patient, from 1998/89 to 2005/06
Figure 4. Trends in rates of availability per 1000 population of public and private beds from 1998/89 to 2005/06
Figure 5. Variation between DHBs in discharge ratios standardised for age, gender, and casemix 2005/06
Limitations of the data—The New Zealand data series provides reasonably reliable comparisons over the period given the filtering, truncation, and standardisation process to which the data is subjected. This ensures that, as far as possible, there is appropriate comparison between years and between DHBs. The principal uncertainties relate to bed availability given the lack of recent data and the inclusion in previous data of beds used for a variety of purposes additional to acute short-term stays. There are also serious uncertainties about private-sector data. None is currently available regarding utilisation. The data, both public and private, is relatively limited compared with other countries, e.g. Australia.
There are also uncertainties about the international comparisons including comparability for population needs, inclusion or exclusion of certain categories of patients excluded from the New Zealand data, e.g. admissions for endoscopies. Despite this the differences are sufficiently marked to indicate that progress in reducing bed utilisation in New Zealand is well ahead of Australia but falls well short of some situations in the US such as Kaiser Permanente.5 This organisation is widely regarded as a benchmark for international studies but its utilisation figures are reasonably typical of those of US-managed care markets.5
Changes in bed utilisation in New Zealand—The New Zealand data clearly points to consistent progress in reducing bed utilisation. Although there has been a continuing increase in discharge rates, almost entirely related to the increase in day patients, there has been a marked fall in the rate of bed days explained almost entirely by the dramatic fall in the ALOS. This is related to a range of factors including an increasing percentage of day surgery, better management of the inpatient episode, new technologies (e.g. endoscopic surgery), and increasing provision of alternative community-based services to enable earlier discharge.
This decrease is in the face of increasing demand for hospital-based procedures including new technologies and drugs and an aging population. The findings clearly indicate that the there has been an increase in cost weighted discharges related to increasing complexity of patients admitted for hospital care. It is almost certain that, despite factors leading to a reducing ALOS, a limit may soon be reached in achieving further reductions in bed utilisation.
Previous studies—The findings of this study are reasonably consistent with previous work in this area. For example, Pool et al reported on a detailed study (age and gender standardised) of hospital bed days over the period 1985–87 to 1999–01.7 Overall rates/1000 reduced from 890 to 520 for males and 720 to 350 for females. Allowing for truncation of the MOH data these rates and trends are similar to this study. However their study did not include trends in rates of discharges and ALOS.7
Katzenellenbogen et al undertook a detailed study of hospital discharge data over a similar period. They concluded that trends in hospital days, appropriately filtered, provided a more reliable picture of utilisation related to health needs than hospital discharges.9 However their work was largely focused on the concept of health expectancy rather than an analysis of trends as such.9
A study of public hospital bed numbers was undertaken by Jackson in 2006.10 He reported a total of 1.66 beds/1000 population, a very similar figure to the calculated 1.56 beds/1000 days in this study.
A study of variation in hospital board utilisation by Barnett reported in 1984 showed that a key factor in hospital utilisation in the past and variation between population areas was highly correlated with bed availability as expressed by Roemer’s Law.11 Malcolm found in 1983 that this relationship was especially marked in mental health services.12
Mansoor studied variation in bed availability and utilisation in the early days of area health boards in 1990.13 He again found wide variation between boards in both rates of admission and bed days. The overwhelming factor explaining this variation, especially in bed days, was bed availability. Need, as measured by the Health and Equity Score used at that time, was not an explanatory factor.
Pool et al also found a wide variation in rates of bed day utilisation between the regional councils used for population comparisons in their study to 2001.7 This variation had markedly decreased over the 15 years of the study period and was very much less that reported in the Barnett and Mansoor studies.11,13 It is almost certain that this is related to a reduction in number of beds in previously overbedded districts and hence the variation in bed availability—a key factor in utilisation variation in the past.
International comparisons—The most obvious explanation for the differences between the four countries/settings presented in Table 2 is the level of integration in health systems especially between primary and secondary care. The DHB system in New Zealand is probably the most integrated of any comparable country perhaps even more than Kaiser in the range of services included. The data from Kaiser Permanente, where the ALOS is similar to that of New Zealand, indicates that much better prospects for further reductions in New Zealand can be achieved by reducing admissions to hospital.
Variation between DHBs—The findings above indicate that even after standardisation for age, gender, and casemix there is still a marked variation between DHBs in discharge ratios in 2006. The highest DHBs have a 36% higher ratio than the lowest. However this variation is very much less than in the past, especially in the days of hospital and area health boards as presented above.
However other factors not included in standardization may be significant. These include deprivation, ethnicity, and rurality as possible explanations for the much higher ratios in Tairawhiti and the West Coast DHBs. However a correlation analysis shows that there are only a small and non-significant relationships between DHB-standardised discharge ratios and mean deprivation score (0.29), % Māori (0.28), and rural status (0.32).
The 16% lower than the mean standardised discharge ratios for Capital and Coast and Mid Central DHBs suggest possible benchmarks for other DHBs to achieve in reducing their discharge rates. It would seem that further research to examine these differences—especially by DHBs with above-average ratios and struggling to keep within their population-based funding, e.g. Canterbury—would be well worth undertaking.
Will we need more or less beds in the future?—It would appear that the main target for reducing bed utilisation and bed numbers in the future is to firmly focus upon reducing hospital admissions. There is no doubt that further gains can be made in increasing the proportion of day patients and reducing ALOS. However the Kaiser data point to the prospects of lower if not very much lower rates of discharges.
Several initiatives have been developed in recent years by IPAs/PHOs to provide alternative community services for potentially admissible patients.14 In general these have been reasonably successful although limited. However there have been few published analyses or even descriptions of these initiatives. A much more comprehensive approach is needed. For example Canterbury DHB is implementing a comprehensive acute demand programme which could have a major impact upon reducing the future need for beds, containing costs, and freeing up resources including for elective surgical services.
Other developing initiatives include PHOs targeting identified patients with chronic diseases such as cardiovascular disease and diabetes.15 A Christchurch study examining the relationship between primary and secondary in 1996/97, on patients registered with the Christchurch South Health Centre showed that patients with chronic diseases and having high use health cards (HUHCs), constituted 8.6% of the population but generated 42.4% of hospital bed days.16
Furthermore, the Christchurch study demonstrated the importance of primary care factors in the utilisation of secondary care especially acute hospitalisation in older patients and the potential for targeting patients in primary care settings to contain hospital admissions.
Given past trends in hospital utilisation and potential for further reductions, Figure 6 shows a possible future for bed availability and utilisation with fuller integration of primary and secondary care. There are questions about the organisational arrangements within DHBs to support this integration.
Figure 6 Possible future trends in bed availability and utilisation in New Zealand with full integration of primary and secondary care
A key problem is the continuing focus upon the hospital as an organisation. The downside of this is that some of those working within this organisation have a strong hospital focused culture that may be threatened by primary and community care alternatives. Despite this, the DHB system has both the potential as well as incentives to realign hospital and community cultures into a more integrated patient focused approach to care.
Competing interests: Elected member of Canterbury District Health Board.
Author information: Laurence Malcolm, Professor Emeritus and Consultant, Aotearoa Health, Lyttelton—and CDHBoard Member, Christchurch
Acknowledgment: I am grateful to Ministry of Health staff for the provision of recent as-yet-unpublished draft data on Hospital Throughput and the Private Surgical Hospitals Association for provision of bed numbers.
Correspondence: Laurence Malcolm, Aotearoa Health, RD1, Lyttelton, North Canterbury. Fax: (03) 329 9084; email: email@example.com
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals