Journal of the New Zealand Medical Association, 23-July-2004, Vol 117 No 1198
THE NEW ZEALAND
Vol 117 No 1198 ISSN 1175 8716
Trends and patterns of avoidable hospitalisations in New Zealand: 1980–1997
Arunachalam Dharmalingam, Ian Pool, Sandra Baxendine, Janet Sceats
New Zealand’s health system has gone through major reforms over the last two decades, particularly in the 1990s.1-3 The health reforms have occurred in the context of major restructuring of the economy and the social sectors.4-5 The direct impact of the health reforms is on the supply of health services: availability of (and accessibility to) primary care. The reforms can influence the level of avoidable hospitalisations by limiting the availability and affordability of primary care.
Evidence from both overseas and New Zealand suggests that there is a strong link between avoidable hospitalisations and the underutilisation of primary care, and that underutilisation is associated with lower socioeconomic status.6–13
Several studies have shown that (with timely and adequate primary care) it is possible to prevent the onset of illness and control an acute episode; chronic conditions can also be monitored and managed with access to appropriate primary care.7,11,12,14-19 In this paper, we describe the trends in (and pattern of) hospitalisations for conditions that could potentially be taken care of by the provision of (and access to) adequate primary care. We describe the regional differentials, age patterns, and disease patterns in avoidable hospitalisations between 1980 and 1997. We then discuss the findings in the context of the changing health sectors of the 1980s and early 1990s.
Data and methods
We use public hospital inpatient discharges data obtained from the New Zealand Health Information Service for the calendar years 1980–1997. To allow for time series analysis, the data were adjusted for changes to data collection and coding procedures and health policy. Katzenellenbogen et al20 have recently documented that a number of health policy and other changes have affected the comparability of discharge data for the period since 1980. They also ‘filtered’ the data by excluding a number of categories, so that valid comparisons could be made over time.21 We have adopted Katzenellenbogen et al’s suggestions, and (accordingly) excluded several categories (including day-patients, as definitions differed over time) from the discharge data set used here.
Several overlapping lists of avoidable hospitalisations are available in the literature. In general, avoidable hospitalisations are defined as those resulting from medical conditions that can be prevented, controlled, monitored, and managed by timely and effective primary care.6,8,12,19 Following Jackson et al,10,11 in this paper we consider the hospitalisations for medical conditions given in Table 1, as avoidable.
Table 1. Avoidable hospitalisation (AH) conditions with ICD-9-CM codes
We derived avoidable hospitalisation rates for each of the four time points around the 1981, 1986, 1991, and 1996 censuses. We first computed a 3-year moving average for the number of avoidable hospitalisations and then divided this figure by the census population to obtain the rates for the four periods: 1980–82, 1985–87, 1990–92, and 1995-97. We computed avoidable hospitalisation rates by gender, functional age groups, and medical condition for New Zealand (as a whole and for 29 geographic regions).
The most recent health reforms have reorganised the New Zealand health sector into 21 District Health Boards (DHB). As some DHBs have a much bigger population or geographical spread than others, for the purpose this paper we divided the bigger ones into smaller units to give a total of 29 regions. Of the 29 regions, 12 regions had a population of less than 100,000 in 1996, 14 regions had a population between 100,000 and 200,000, and 3 regions had a population over 200,000 people in 1996.
All the rates presented here are standardised for age compositional differences, using the 1996 New Zealand age-composition as the standard. This is important because the demographic composition of a population can not only change over time but can also differ between various geographic regions.
To study the regional differences in avoidable hospitalisations, we derived relative hospitalisation rates for each region by using North Shore as the reference population. North Shore was chosen as it had one of the lowest avoidable hospitalisation rates among the 29 regions. North Shore is also a region that has fared well in many socioeconomic indicators. Relative avoidable hospitalisation rate is obtained by dividing the age-standardised avoidable hospitalisation rate for a given region and year by the age-standardised avoidable hospitalisation rate for North Shore for the same year.
We use the proportion of sole Maori population (the indigenous people of New Zealand) in a region as a proxy for socioeconomic status. There is a large body of empirical literature that shows that Maori ethnicity is strongly associated with poor socioeconomic status.22-27
In New Zealand censuses since 1986, the ethnicity question has been based on self-identification, and individuals can report more than one ethnicity. Sole Maori are those who identify Maori as their only ethnic group. Until the 1981 census, ethnicity classification was based on a biological criterion: all those with half or more Maori blood were classified as Maori.
Moreover, until 1996 people could report only one ethnicity in hospitalisation records. The inconsistencies, in the definition and collection of ethnicity data between the censuses (that provide the denominator data) and hospitalisation records, could lead to underestimation or overestimation of hospitalisation rates among the Maori—depending on whether sole Maori or mixed Maori population is used in the denominator.28-30 Thus we decided not to derive hospitalisation rates by ethnicity.
We also considered the proportion of population receiving welfare benefits in a region and the proportions without any educational qualifications as regional level socioeconomic indicators. But this information was not comparable over the time period under consideration. For instance, ACC was only included in the benefit data from 1991; ‘no educational qualifications’ data was also problematic because there was a ‘still at school category’ in 1981, 1986, and 1991 censuses—but not in the 1996 census.
We did not have access to the NZdep data (aggregate area level composite measure of deprivation derived from the census based on a number of socioeconomic indicators) but again they are available only for 1991 and 1996 censuses. Thus we decided to use the proportion of sole Maori as an indicator for the relative socioeconomic position of a region. The correlation analysis (results not given) showed that the proportion of sole Maori in a region was strongly associated with the proportion on welfare benefits (for all but 1980–82 period) and the proportion with no educational qualifications.
Although the ethnicity questions in the census have changed over time, data on sole Maori obtained since the 1986 census (based on sociocultural definition) are comparable with the data from the 1981 census (based on blood-fraction definition).28,29
Age standardised avoidable hospitalisation rates (AHR) are given in Table 2 for the period 1980–97 by geographic regions. For New Zealand (as a whole), the rate increased from 73 per 10,000 people in 1980–82, to 83 in 1985–87, and to 96 in 1995–97.
Table 2. Age-standardised avoidable hospitalisation rates per 10,000 population in New Zealand
This amounts to an increase of about 32% between 1980 and 1997. This increase was punctuated by a decline between 1985–87 and 1990–92. The rates for males and females followed a similar trend. Males have a higher AHR than females, and the male-female difference seems to have narrowed: the difference declined from about 16 per 10,000 people in the 1980s to about 14 in 1995–97. There was a substantial geographical variation in the level of avoidable hospitalisation—over the 1980–97 period, some regions experienced greater increases in AHRs than others.
While most regions showed decline in the rate of hospitalisations between 1985–87 and 1990–92, nine regions showed monotonic increase during 1980–97. These regions are: Northland, Central Waikato, Eastern Bay of Plenty, Lakes, Tairawhiti, Wanganui, Southland, Rodney, and Waitakere. Incidentally, all but Southland, Rodney and Waitakere have a significant rural and sole Maori population. Tairawhiti, a region with over one-third of its population in 1991 being sole Maori, had the highest rate of avoidable hospitalisations since the mid-1980s.
The relative avoidable hospitalisation rates (RAHR) for the 29 regions are given in Table 3. As North Shore had one of the lowest avoidable hospitalisation rates since 1980, it was chosen as the reference region to derive the RAHRs. The pattern and trends in the relative rates help illustrate how the different regions have experienced avoidable hospitalisations relative to a relatively better off region, North Shore.
It is clear from Table 3 that in 1980–82 there were about 13 regions whose rates were over two times as much as the rate observed for North Shore. This declined to six regions in 1985–87, to four regions in 1990–92, and to only one region in 1995–97. This underscores the fact that (over time) the various regions have tended to converge towards the rates observed in North Shore.
There is also a strong socioeconomic gradient to the regional pattern in avoidable hospitalisations. In general, regions with higher avoidable hospitalisation rates have tended to have a greater proportion of sole Maori populations. In 1995-97, Eastern Bay of Plenty and Tairawhiti were the only regions that had rates about twice the rate observed in North Shore. These two regions also had the highest proportion of sole Maori population (about 35% in 1991 census). This is further supported by the simple regression estimates given in Table 4: the proportion of Sole Maori population in a region is a powerful predictor of the avoidable hospitalisation rates.
In the 1990s, over 80% of the regional variations in avoidable hospitalisation rate is explained by the regional variations in the proportion of sole Maori populations. We also considered including two additional variables in the regression model: proportion of population on benefits (unemployment, domestic purposes benefit (DPB), and sickness/invalid benefit) and proportion with no educational qualifications. But we did not include them because, as discussed above, the data on these two indicators were not comparable for the periods. Moreover, these two variables were very strongly associated with proportion of sole Maori population. This association suggests that ethnicity is probably a reliable marker of socioeconomic deprivation, including inequalities in education and employment.
Table 3. Relative avoidable hospitalisations in New Zealand (relative to North Shore region)
Age-standardised total hospitalisation rates (including avoidable and non-avoidable hospitalisations) showed that (tables not included) there was an overall decline between 1980–82 and 1995–97—although there was a slight increase between 1990–92 and 1995–97. The time trend in total hospitalisation rates is similar to that of avoidable hospitalisation rates given in Tables 2 and 3, but in the opposite direction. Again there was some regional variation.
Table 4. Parameter estimates from a simple regression model for four periods in New Zealand
Note: Age-standardised avoidable hospitalisation rate is the dependent variable; proportion Sole Maori is the independent variable.
Although almost all regions (see Table 2) had experienced an increase in avoidable hospitalisations between 1980–82 and 1995–97, all but ten regions had a decrease in total hospitalisation rates during the same period. Thus overall, the avoidable hospitalisation rates and total hospitalisation rates moved in the opposite direction during 1980–97. As with the relative avoidable hospitalisation rates, the variations in relative total hospitalisation rates have tended to narrow over time. Thus a major change in the level of total and avoidable hospitalisation rates is for regional variations to narrow.
In Table 5, we provide the avoidable hospitalisation rates as a percent of total hospitalisation rates for various regions. In 1995–97, about 1 in 10 hospitalisations were avoidable. For New Zealand (as a whole), the avoidable as a percent of total hospitalisations has increased from 7% in 1980–82 to 10%in 1995-97. Interestingly, there is no clear socioeconomic gradient to the observed pattern in Table 4. This is likely to be related to the regional variation in the direction of change in total and avoidable hospitalisation rates over the time period (1980–97).
Rates of avoidable hospitalisations for seven broad age groups are given in Table 6. The highest rate is observed for those aged 75+, followed by the youngest population aged under 5 years. In 1995–97, the avoidable hospitalisation rates were 409 per 10,000 population in the 75+ age group, 228 in the under 5 age group, 180 in the 65–74 age group, and between 45 and 75 in the 5–64 age groups.
As with the overall avoidable hospitalisation rates, age-specific rates also showed an increase between 1980–82 and 1995–97. However, the rate of increase differed by age. While the rate increased by 31% between 1980–82 and 1995–97 for the under 5 age group, it did so by 51% among those aged 75 and over. The smallest increase (2%) was observed for the 5–14 age group.
Table 5. Percentage of total hospitalisations that are avoidable in New Zealand (for various regions)
Socioeconomic variations are also evident in the age-specific rates for all the periods under consideration. But socioeconomic differences seem to vary by age. For instance, in 1995-97, children aged under 5 years in regions with over 15% sole Maori population were about 37% more likely to experience avoidable hospitalisations compared to children in regions with less than 10% sole Maori population (see Figure 1). But among those aged 75+, they were only 17% more likely to experience avoidable hospitalisation if they lived in a region with over 15% sole Maori population than those living in a region with under 10% sole Maori population.
Table 6. Age-specific avoidable hospitalisation rates in New Zealand (per 10,000 people)
The avoidable hospitalisation rates as a proportion of total hospitalisation rates for different age groups are given in Table 7. It is interesting that the two youngest age groups (under 5, and 5–14) had the highest levels: over 12% of all hospitalisations was avoidable. The 75+ age group was the next highest. Thus not only that the level of avoidable hospitalisation was one of the highest among children aged under 5 years, but also the rate was higher in relation to total hospitalisation.
Table 7. Percent of total hospitalisations that are avoidable in New Zealand (for various age groups)
In Table 8 we present the avoidable hospitalisation rates for a number of primary care conditions (see Table 1 for a list of conditions). We have followed Jackson et al10 in classifying various diagnoses into a number of primary conditions.6 It is clear from Table 8 that people were more likely to be hospitalised for asthma, pneumonia, congestive heart failure, and cellulitis than for any other avoidable conditions.
Table 8. Age-standardised avoidable hospitalisation rates for various primary care conditions in New Zealand
In 1995–97, the avoidable hospitalisation rates for asthma and pneumonia were the highest (over 250 per 100,000 people), followed by congestive heart failure (173), and cellulitis (155). In terms of time trend, of the four leading conditions of avoidable hospitalisation, all but asthma showed a consistent increase between 1980–82 and 1995–97. In fact, the increase was particularly dramatic between 1990–92 and 1995–96: avoidable hospitalisation rate for cellulitis (a predominantly adult condition) increased by 70%, by 62% for pneumonia and by 21% for congestive heart failure.
On the other hand, the rate for asthma decreased by 29% between 1985-97 and 1995-97; as did the rate for immunisable conditions (by 34% between 1985–87 and 1995–97), which (along with asthma) occur predominantly among children.
Figure 2 shows the pattern of avoidable hospitalisation for selected primary care conditions for three groups of regions—those with under 10% sole Maori, between 10% and 14%, and those with over 15% sole Maori.
In general, for the four conditions shown in Figure 2, the higher concentration of sole Maori populations is associated with higher likelihood of avoidable hospitalisation. This was true for all the four time points. What is also interesting to note from Figure 2 is that when we compare the rates for asthma and pneumonia for the periods 1990–92 and 1995–97, while the socioeconomic differential seemed to narrow for asthma between 1990-92 and 1995-97, it seems to have increased for pneumonia.
Figure 1. Age standardised avoidable hospitalisation rates for four major conditions for three groups of regions in New Zealand (Sole Maori under 10%, 10-14%, and 15+%)
Figure 2. Age standardised avoidable hospitalisation rates for four major conditions for three groups of regions in New Zealand (sole Maori under 10%, 10-14%, and 15+%)
The analysis has shown that the rate of avoidable hospitalisations has increased between the early 1980s and mid-1990s. But there was a decline in avoidable hospitalisations between the mid- and late-1980s. This period coincided with the first major health reforms (1983–1991). Fourteen locally elected Area Health Boards were formed during this period with the aim of improving efficiency and accountability. The boards were given the responsibilities for health promotion, disease prevention, and provision of personal treatment and caring services.2,31
Although, it was criticised that there was no incentive for the health boards to be efficient and that they suffered from weak accountability,32 some area health boards did make efforts to strengthen the primary care system. In order to improve access to primary care, subsidies to general practitioner services were increased.3 The decrease in avoidable hospitalisation between 1985–87 and 1990–92 might have been partly due to the changes effected by primary care initiatives of the area health boards.
The increase in avoidable hospitalisations between 1990–92 and 1995–97 is likely to be associated with the radical health reforms introduced in 1991 but came into effect in 1993.1-3,11,33-36 The main features of the 1993 reforms included: the separation of the funding, purchasing and provision of health services with a view to encourage competition, and market behaviour in the health sector.
Although there was no formal evaluation of the impact of the reforms on the utilisation of health services, it has been argued that the reforms were likely to have had the deleterious effect of severely limiting access to primary care. This was not only due to the increases in fees for general practitioner services but also due to welfare cuts which affected the ability to access health services.37,38 This was the case at least until the end of 1996 when the new Coalition Government significantly modified the 1993 reforms.33,39-41
Although community service cards were introduced for low-income groups as part of the health reforms to enable access to healthcare, recent studies showed that the disadvantaged groups are still likely to underutilise primary care.42-47
Another interesting result was the narrowing of the regional disparities in avoidable hospitalisations between 1980 and 1997. Perhaps this is linked to the change over to population-based funding for the provision of health services beginning with the introduction of area health boards in the mid-1980s. This formula is based on the age/gender composition of a health area and is adjusted for the proportion of low-income households.48 The reduction in regional inequalities in the levels of hospitalisations (both avoidable and non-avoidable) could be one of the beneficial impacts of the health reforms.
Although the proportion of sole Maori population in a region is not necessarily a perfect measure of relative deprivation of a region (as a reviewer pointed out, this factor is likely to be confounded in the analysis with the number of GPs in a region), our analysis has shown that it is a good predictor of the level of avoidable hospitalisations. It is well established that regions with high Maori population are more likely to have higher proportion with no educational qualification, higher level of unemployment, poor housing, higher proportion of sole parents, and higher proportion on welfare benefits.49
Several studies have also shown that poor socioeconomic status is associated with underutilisation of primary care.1,10,11,16,47,50-53 Although financial barriers are critical, non-monetary factors such as transport and geography may also play a role in the poor utilisation of primary care.17
It is suggested that the ‘lack of purpose in life’ among the less educated and unemployed could be factor for poor utilisation.54,17 It may be necessary for the primary health care professionals to develop trusting relationship with the patients to motivate those who have lost purpose in life. However, developing a trusting relationship is difficult when faced with a higher patient to doctor ratios.54,43
Moreover, it is argued that the primary care clinicians ‘do not use community-based information to organise the content of the medical care they deliver’; rather ‘they often use practice-based data to plan for the delivery of services’.46 Access to primary care is also limited by the uneven distribution of general practitioners across the geographical regions. Despite substantial increases in the number of general practitioners in New Zealand, maldistribution continues to be a perennial issue.42
Since the 1993 reforms, there have been two re-reforms in 1996 and 1999.2 The late 1990s reforms have tried to allow greater community participation in health sector decision-making and replace ‘competition’ with ‘collaboration’.1,2,33,34,39-41 The ‘re-reforms’ (underpinned by greater emphasis on equity, social justice, and community voice in decision-making) have the potential to reduce the cost of healthcare and improve the quality of life by increasing access to good primary care and reducing avoidable hospitalisations.
Author information: Arunachalam Dharmalingam, Senior Lecturer, Population Studies, Department of Sociology and Social Policy, University of Waikato, Hamilton; Ian Pool, Professor of Demography, Population Studies Centre, University of Waikato, Hamilton; Sandra Baxendine, Research Fellow, Population Studies Centre, University of Waikato, Hamilton; Janet Sceats, Managing Director, Portal Consulting and Associates, Hamilton.
Correspondence: A. Dharmalingam, Population Studies, Department of Sociology and Social Policy, University of Waikato, Private Bag 3105, Hamilton. Fax: (07) 838 4654; email: email@example.com
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals