Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Patient-initiated switching between private and public inpatient hospitalisation in Western Australia 1980 – 2001: An analysis using linked data

Patient-initiated switching between private and public inpatient hospitalisation in Western... Background: The aim of the study was to identify any distinct behavioural patterns in switching between public and privately insured payment classifications between successive episodes of inpatient care within Western Australia between 1980 and 2001 using a novel 'couplet' method of analysing longitudinal data. Methods: The WA Data Linkage System was used to extract all hospital morbidity records from 1980 to 2001. For each individual, episodes of hospitalisation were paired into couplets, which were classified according to the sequential combination of public and privately insured episodes. Behavioural patterns were analysed using the mean intra-couplet interval and proportion of discordant couplets in each year. Results: Discordant couplets were consistently associated with the longest intra-couplet intervals (ratio to the average annual mean interval being 1.35), while the shortest intra-couplet intervals were associated with public concordant couplets (0.5). Overall, privately insured patients were more likely to switch payment classification at their next admission compared with public patients (the average rate of loss across all age groups being 0.55% and 2.16% respectively). The rate of loss from the privately insured payment classification was inversely associated with time between episodes (2.49% for intervals of 0 to 13 years and 0.83% for intervals of 14 to 21 years). In all age groups, the average rate of loss from the privately insured payment classification was greater between 1981 and 1990 compared with that between 1991 and 2001 (3.45% and 3.10% per year respectively). Conclusion: A small but statistically significant reduction in rate of switching away from PHI over the latter period of observation indicated that health care policies encouraging uptake of PHI implemented in the 1990s by the federal government had some of their intended impact on behaviour. Page 1 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 the Health Insurance Act of 1973. Thus switches between Background Coexistence of public and private health insurance, such use of the public and PHI system are initiated solely by as in Australia, has been the subject of intense debate patients based on choice rather than being mandated by among health economists and policy makers [1]. The government, notwithstanding that this 'choice' may be main issue surrounding this debate has been how the mix constrained in some instances by socioeconomic and of public and private health care financing influences the locational access factors. We propose that since posses- demand for private health insurance (PHI) and whether sion and utilisation of PHI are not equivalent, analysing PHI takes pressure off the public system [2]. the effectiveness of the recent government strategies in relieving the pressure on the public system cannot be Falling PHI membership, observed since the introduction accomplished by evaluations of changes in possession of of Medicare in 1984, was thought to have increased the PHI alone. Rather, changes in choice, as reflected by demand on the public system [3], prompting the federal patient-initiated switching between the public and private government to implement policies aimed at encouraging insurance systems must be analysed. possession of PHI to take the pressure off public hospitals and restore balance to the health care system [3-5]. Since The aim of this study was to identify and measure changes 1995 three major policy reforms have been introduced in in the behavioural patterns of switching between public Australia [6]. Firstly, in 1995 selective contracting was and privately insured status for hospitalisation by the introduced. The then federal Labor government passed population of Western Australia using our novel couplet legislation allowing private health plans to contract selec- methodology for analysing longitudinal data. tively with hospitals and doctors so as to improve compe- tition on price and quality. Secondly, government Method subsidisation of PHI was introduced in 1997 by the Con- Hospital morbidity data extraction and case selection servative coalition federal government as a means-tested The WA Data Linkage System [13] was used to extract all st rebate, capped at a flat amount irrespective of the cost of hospital morbidity data system (HMDS) records from 1 st PHI, combined with an income tax surcharge for high January 1980 to 31 December 2001, containing income earners without PHI. The rebate part of the policy encrypted patient identification and episode numbers, was subsequently replaced in 1999 by a non-means-tested age, gender, date of admission, date of separation, pay- 30% rebate for PHI available to everyone. Finally, in ment classification (public, insured private, or "other"), 2000, lifetime community rating was introduced. This and hospital type. The "other" payment category, which policy relaxed the previous stringent community rating included the private uninsured (2.2% of the total epi- system by allowing the price of PHI to be varied according sodes), workers compensation (1.8%), motor vehicle to the age at which a member joined [6]. (0.7%), defence force personnel (0.3%) and Veteran Affairs (1.7%) classifications was removed from the data The Australian Healthcare Agreement 1998–2003 com- set, leaving only the categories of public and private mitted the Commonwealth and states to review the rela- insured. This was done because the study was concerned tionship between PHI cover and the use of hospital with elective shifts between PHI and public categories; not services by private patients. The investigation of this rela- enforced payment classifications due to mandatory fund- tionship is becoming a priority as there is disagreement ing arrangements, or private episodes for which the among commentators as to the financial efficiency of the patients paid the full cost. 30% rebate [5,7-10]. To date, analyses of the effects of pol- icies aimed at supporting PHI in Australia have primarily Data coverage centred on changes in the proportion of the population This research has made use of records from the HMDS covered by PHI [3,6,11,12]. However, changing the pro- which is the inpatient information system for WA acute portion of the population covered may not directly trans- care hospitals. The data collected by the HMDS are patient late to increased utilisation and, therefore, reduce pressure identification, socio-demographic, services, administra- on the public system. The relationship between PHI cover tion and clinical diagnosis information. Every WA hospi- and type of hospital use is complex because the universal- tal defined as an acute care facility has to provide data via ity of Medicare in Australia means that anyone can be the Health Act or for private hospitals as part of their treated in a public hospital at no charge regardless of their license and every free-standing day surgery unit must also insurance status. provide data. The HMDS includes patient information for all acute A class hospitals, day surgery units, geriatric and The right of individuals to choose between public and pri- psychiatric inpatient facilities in general hospitals and, as vate insurance, regardless of the status (public or privately of July 1993, healthy newborn infants born in hospital. It insured) of previous hospitalisations or the possession of does not include patients of stand-alone geriatric and psy- PHI is protected by the principles of Medicare as set out in chiatric institutions, geriatric hostels or rehabilitation Page 2 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Couplet 1 Couplet 2 Episode of Hospitalisation Individual Patient Record 1st Episode 2nd Episode 3rd Episode Th Figure 1 e formation of hospital couplets The formation of hospital couplets. units; however, most metropolitan hospitals have geriat- Stratification by age group ric and psychiatric units attached for which data are pro- Stratification was based on age at admission of the first vided for the HMDS. Over the lifetime of this study there episode in each couplet. Each episode of care was assigned were no significant changes in the HMDS coverage. to one of four broad age categories (0–16 years, 17–39 years, 40–69 years and 70+ years) chosen to represent PHI Formation of episodes of care and assignment of payment market segmentation (children, young adults, middle age classification and old age) following consultation with a local private For each individual in the data set all eligible hospital health insurer. records were grouped into episodes of care, using the sep- Formation and classification of hospital couplets aration and admission dates to define temporally contig- uous periods of health care service utilisation. Thus one For each individual in the data set, eligible episodes of episode of care could have contained one or more inter care were grouped incrementally, starting with the index hospital transfers. Each episode of care was assigned to (first episode of care) to form hospital couplets such that one of the two eligible payment classifications (public or episodes 1 and 2 formed hospital couplet 1, episodes 2 privately insured) on the basis of the initial payment clas- and 3 formed hospital couplet 2 and so on (see figure 1). sification at admission, where inter-hospital transfers Hospital couplets were categorised depending upon the were involved. Allocation of public or PHI status was sequential combination of payment classifications of based solely on payment classification and not on hospi- their two contributing episodes of care as follows: tal type. This was done because the focus of this paper (consistent with the focus of public policy making in Aus- (i) Concordant public couplets containing only public tralia) was on use of PHI rather than hospital type. episodes of care. Page 3 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 (ii) Concordant private couplets containing only privately recorded for these individuals formed 6,376,472 distinct insured episodes of care. hospital couplets (see table 1). Of these, approximately one half (51%) had an intra-couplet interval of one year (iii) Discordant (mixed) public to private couplets had or greater. Details of the distribution of the hospital cou- the first episode of care as public and the second as pri- plets by intra-couplet interval are shown in table 2. Signif- vately insured. icant differences were observed in the characteristics of couplets having intra-couplet intervals of less than 1 year (iv) Discordant (mixed) private to public couplets had the compared with those of one year or over. The largest dif- first episode of care as privately insured and the second as ferences were observed in relation to hospital type and public. couplet category. Analysis of the mean intra-couplet interval The distribution of first couplet episodes in each year of The intra-couplet interval was defined as the time in days observation was more uniform in couplets with less than between the final separation from the first episode of care a one year intra-couplet interval compared with couplets and admission to the second episode of care. The mean longer intra-couplet interval or all couplets in the data file intra-couplet interval was calculated for each couplet clas- as shown in figure 2(A). The reduction of first couplet epi- sification by the year of admission of the second episode sodes in couplets with intervals = one year was directly of care (a term we subsequently refer to as the couplet proportional to the number of years remaining in which a year) and expressed as a ratio of the sum of the mean second episode (thus completing a couplet) could be intra-couplet interval of all couplets in that year, regard- observed. The lack of first episodes observed in 1980 was less of classification. We subsequently refer to this meas- the result of a reduced volume of data in the original ure as the grand mean. Thus a ratio greater than one for a HMDS file for that year, most likely caused by an extrac- particular couplet classification was indicative of an intra- tion error. couplet interval longer than the grand mean for that cou- plet year. The couplet year indicates the year the switch Figure 2(B) shows the distribution of second episodes in (choice) was made. couplets over the observation period. The proportion of second episodes increased over the observation period Analysis of behavioural patterns in switching of payment regardless of the duration of the intra-couplet interval. classification This was a function of the increased number of individu- The proportion of each type of discordant couplet relative als eligible to complete a couplet with a second episode as to the total number of couplets having a first episode of time progressed. care in the baseline payment classification was deter- mined independently for all intra-couplet intervals (in Mean intra-couplet interval whole years) in the data set. The ratio of the mean intra-couplet interval observed for each couplet category relative to the grand mean by age This analysis was performed separately for each age group group and couplet year is shown in figure 3. Discordant for the whole observation period and two predefined time couplets had the longest intra-couplet intervals, having on periods (1981 to 1989 and 1990 to 2001) chosen to rep- average a ratio relative to the annual grand mean intra resent the two main eras of health care policy in Australia. couplet interval of 1.35, while concordant couplets types The first related to the removal and re-introduction of free had the shortest intra-couplet intervals, their ratio being public hospital care, while the second related to changes 0.65. The overall pattern indicated that the longer the time in federal health policies aimed at supporting PHI between the first and second episode of a couplet, the [5,12,14,15]. Hospital couplets were partitioned into the more likelihood there was of a change in payment classi- two time periods using the year of admission of second fication, especially where the first payment classification episodes of care. The values obtained were plotted as seg- was private. The trends also indicated that, within each mented trend lines and the average rates of loss from each age group, individuals with public concordant couplets, payment classification per year of intra-couplet interval on average, had shorter intervals between episodes (ratio were calculated using least squares fit. 0.5) than individuals with private concordant couplets (ratio 0.8). Results Characteristics of the hospital couplets in the HMDS data The removal / re-introduction of free public hospital care file (1980 – 1984) has previously been shown to have had a The HMDS data file contained data pertaining to significant effect on use of in-patient insurance classifica- 1,979,946 individuals of which 1,185,014 (60%) had at tions [16]. In this study no significant difference was least one valid couplet. The 7,561,486 episodes of care observed with regard to when the first episode of a couplet Page 4 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Table 1: Characteristics of individuals, hospital episodes and hospital couplets nd Characteristic Individuals Episodes Couplets (2 Episode) Number % of Dataset Number % of Dataset Number % of Dataset Sex Male 522167 44.1 3156438 41.7 2634271 41.3 Female 662841 55.9 4405035 58.3 3742194 58.7 Indeterminate 4 0.0 13 0.0 7 0.0 TOTAL 1185014 100 7561486 100 6376472 100 Age Group 0–16 Years 313244 26.4 1045092 13.9 731848 11.5 17–39 Years 443602 37.4 2495039 33.1 2051437 32.3 40–69 Years 326974 27.6 2708572 35.9 2381598 37.5 70+ Years 100870 8.5 1292462 17.1 1191592 18.7 1 2 3 7541165 100 6356475 100 TOTAL 1184690 100 Hospital Teaching 367964 31.1 2852257 37.7 2484293 39.0 Type Public Metropolitan 198456 16.7 902286 11.9 703830 11.0 Private Metropolitan 313563 26.5 1838354 24.3 1524791 23.9 Public Country 279315 23.6 1745527 23.1 1466212 23.0 Private Country 25097 2.1 208883 2.8 183786 2.9 4 5 6 TOTAL 1184395 100 7547307 100 6362912 100 324 missing 20321 missing 19997 missing 619 missing 14179 missing 13560 missing Missing records were caused by missing data in the relevant fields of the individual records and do not reflect deficiencies in the linkage process. occurred (pre or post Medicare); however, the point in Behavioural patterns in switching of payment time of the second episode did correspond with behav- classification ioural change as can be seen in figure 3. Differences were observed in the rate of loss from the pub- lic and privately insured payment classifications across Differences in trend were observed across the four age age groups as shown in table 3. Across all age groups the groups. The 0–16 years age group trended towards a largest losses, at all intra-couplet intervals, occurred from reduction in the intra-couplet interval associated with dis- the privately insured payment classification. The largest cordant private to public couplets after 1985. This was not differences in the rates of loss were observed in the 70+ observed during the late 1980s and early 1990s in the years and the 17–39 years age groups, where an additional other three age groups. The 40–69 years age group showed 2.39 and 1.78 percent of privately insured episodes, a trend pattern similar to the 17–39 years age group until respectively, were lost for every year of intra-couplet inter- 2000, when the average intra-couplet interval associated val. The rates of loss from the privately insured payment with discordant private to public couplets reduced classification over shorter intra-couplet intervals (defined sharply. In addition there was a sharp increase in the aver- as 0 to 13 years) were greater than the rates of loss over age intra-couplet interval associated with discordant pub- longer intra-couplet intervals (defined as greater than 14 lic to private couplets (also observed in the 0 to 16 years years) in all age groups with the largest difference being age group) observed at that time. In the oldest age group observed in the 40–69 years age group and the smallest in (70+ years) the average difference in intra-couplet interval the 70+ years age group. The definition of short and long between public and private concordant couplets was intra-couplet interval was based on an observed substan- much smaller than observed in any of the other three age tial change of slope (inflection) in the segmented trend groups. lines. Over all age groups the trend in concordant couplets was Figure 4 shows the degree of switching from private to that of a slowly reducing average intra-couplet interval, public episodes over the two designated eras in health care excluding the 1983 – 1984 period, where the average policy. In all age groups the average rate of switching away intra-couplet interval for private concordant couplets from the private sector in 1991–2001 (era 2) was lower increased in all age groups. than that observed in 1981–1990 (era 1). The decrease in Page 5 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Table 2: Distribution of hospital couplets by characteristics and intra-couplet interval Characteristic Intra-Couplet Interval (I-CI) Less than 1 Year 1 Year or greater Number % in Dataset % Across I-CI Number % in Dataset % Across I-CI Sex Male 1397957 44.6 53.1* 1236314 38.2 46.9 Female 1739643 55.4 46.5* 2002551 61.8 53.5 Indeterminate 5 0.0 71.4 2 0.0 28.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 Age Group 0–16 Years 318255 10.2 43.5* 413593 12.8 56.5 17–39 Years 871234 27.9 42.5* 1180203 36.5 57.5 40–69 Years 1271495 40.7 53.4* 1110103 34.4 46.6 70+ Years 664439 21.3 55.8* 527153 16.3 44.2 1 2 TOTAL 3125423 100 49.2* 3231052 100 50.8 Hospital Type Teaching 1554708 49.6 62.6* 929585 28.7 37.4 Public Metropolitan 249979 8.0 35.5* 453851 14.0 64.5 Private Metropolitan 573729 18.3 37.6* 951062 29.4 62.4 Public Country 672844 21.4 45.9* 793368 24.5 54.1 Private Country 77205 2.5 42.0* 106581 3.3 58.0 Other 9140 0.3 67.4* 4420 0.1 32.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 Couplet Category Concordant Public 2104333 67.1 56.7* 1605368 49.6 43.3 Private 887622 28.3 43.0* 1176629 36.3 57.0 Discordant Public – Private 74132 2.4 31.5* 161140 5.0 68.5 Private – Public 71518 2.3 19.4* 296730 9.1 80.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 * Statistically significantly different (at the 0.05% level) to the proportion of couplets with an intra-couplet interval 1 year or greater. 12182 missing 7815 missing rate was small (average across all age groups -0.35% per in their next admission than public patients, irrespective intra-couplet year) but statistically significant. In addi- of the length of time between the two episodes. There are tion, significance testing of the difference between each a number of possible explanations for this including puta- pair of proportions (era 1 versus era 2) indicated a signif- tive structural and cognitive reasons for the observed icant difference for the majority as indicated in figure 4. behaviour. Structurally, the average patient who begins with a public classification is likely to be of lesser socioe- We also found that as the intra-couplet interval increased, conomic means than the average patient who begins with the difference between the proportions of discordant cou- a private insured classification. The option of the former plets having a private first episode versus a public first epi- patient to use PHI at the next hospitalisation will be sode increased from 4.2% at one year to 34.3% at 18 years dependant upon them taking out, or at the very least (data not shown). This indicated that overall, regardless of maintaining (if they had private cover at the initial epi- age, there was greater switching away from PHI than away sode but did not use it) private cover in the meantime. The from the public classification. same pre-requisite does not exist for an initially privately insured patient accessing a public classification on the next occasion. Also patients with private insurance experi- Discussion As expected, due to their greater capacity to move between encing trauma or an acute disease event may, in some cir- respective payment classifications, private patients were cumstances, be admitted in an emergency as a public found to be more likely to switch payment classifications patient, thus to some extent disallowing the patient from Page 6 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 14 14 14 14% % % % 12 12 12 12% % % % 10 10 10 10% % % % 8% 8% 8% 8% st st %o %off 11 Ep Epiis sode odes s 6% 6% 6% 6% in in Co Cou up plet lets s 4% 4% 4% 4% 2% 2% 2% 2% 0% 0% 0% 0% st st Y Year ear o off 1 1 C Coup ouplle ett E Epi pis so ode de 10 10 10 10 10 10 10 10% % % % % % % % Al Al Al Al Al Al Allllllll C C C C C C Co o o o o o ou u u u u u up p p p p p pllllllle e e e e e et t t t t t ts s s s s s s > >= = 1 1 1 1 1 Y Y Y Y Ye e e e ea a a a ar r r r r IIIIInt nt nt nt ntr r r r ra a a a a - - - - ---C C C C C C Cou ou ou ou ou o ou up p p p p p pllllllle e e e e et et t t t t t IIIIIIInt nt nt nt nt n nte te e e e e er r r r r r rv v v v v v val al a a a a alllll 8% 8% 8% 8% 8% 8% 8% 8% < < < < < < < 1 1 1 1 1 1 1 Y Y Y Y Y Y Ye e e e e e ea a a a a a ar r r r r r r In In In In In In Intttttttr r r r r r ra a a a a a a -------C C C C C C Co o o o o o ou u u u u u up p p p p p ple le le le le le lettttttt IIIIIIIn n n n n n nttttttte e e e e e er r r r r r rv v v v v v va a a a a a alllllll %o %o %o %o %offfff 2 2 2 2 2n n n n nd d d d d 6% 6% 6% 6% 6% 6% 6% 6% Epi Epi Epi Epi Epis s s s sod od od od ode e e e es s s s s in in in in in C C C C Co o o o ou u u u up p p p pllllle e e e ets ts ts ts ts 4% 4% 4% 4% 4% 4% 4% 4% 2% 2% 2% 2% 2% 2% 2% 2% 0% 0% 0% 0% 0% 0% 0% 0% Distribution Figure 2 of the proportion of first and second episodes in couplets by year Distribution of the proportion of first and second episodes in couplets by year. Page 7 of 11 (page number not for citation purposes) 198 198 1980 0 0 19 19 19 19 19 19 19 198 8 8 8 8 8 8 80 0 0 0 0 0 0 0 19 19 1982 82 82 19 19 19 19 19 19 19 198 8 8 8 8 8 8 82 2 2 2 2 2 2 2 19 19 1984 84 84 19 19 19 19 19 19 19 198 8 8 8 8 8 8 84 4 4 4 4 4 4 4 19 19 1986 86 86 19 19 19 19 19 19 19 198 8 8 8 8 8 8 86 6 6 6 6 6 6 6 1 1 1988 988 988 19 19 19 19 19 19 19 198 8 8 8 8 8 8 88 8 8 8 8 8 8 8 1 1 1990 990 990 19 19 19 19 19 19 19 199 9 9 9 9 9 9 90 0 0 0 0 0 0 0 1 1 1992 992 992 19 19 19 19 19 19 19 199 9 9 9 9 9 9 92 2 2 2 2 2 2 2 1 1 1994 994 994 1 1 1 1 1 1 1 1994 994 994 994 994 994 994 994 19 19 1996 96 96 1 1 1 1 1 1 1 1996 996 996 996 996 996 996 996 1 1 1998 998 998 19 19 19 19 19 19 19 199 9 9 9 9 9 9 98 8 8 8 8 8 8 8 200 200 2000 0 0 20 20 20 20 20 20 20 200 0 0 0 0 0 0 00 0 0 0 0 0 0 0 Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 0- 0- 0-16 16 16 Y Y Year ear ears s s 17 17--3 39 9 Y Yea earrs s 2.0 2.0 2.0 2. 2. 2. 2.0 0 0 0 1.6 1.6 1.6 1. 1. 1. 1.6 6 6 6 1.2 1.2 1.2 1. 1. 1. 1.2 2 2 2 Ra Ra Ratttiiio o o = = = 1 1 1 Ra Ratio tio = = 1 1 0.8 0.8 0.8 0. 0. 0. 0.8 8 8 8 0.4 0.4 0.4 0. 0. 0. 0.4 4 4 4 0 0 0 0 0 0 0 4 4 40 0 0-6 -6 -69 9 9 Y Y Ye e ea a ars rs rs 7 70+ 0+ Y Year ears s 2. 2. 2. 2.0 0 0 0 2.0 2.0 2.0 1. 1. 1. 1.6 6 6 6 1.6 1.6 1.6 1. 1. 1. 1.2 2 2 2 1.2 1.2 1.2 Ra Ra Ratttiiio o o = = = 1 1 1 Ra Ratio tio = = 1 1 0. 0. 0. 0.8 8 8 8 0.8 0.8 0.8 0. 0. 0. 0.4 4 4 4 0.4 0.4 0.4 0 0 0 0 0 0 0 Publ Publiic c C Conc oncord ordant ant P Prriiv vat ate e C Conc onco orrd dant ant M Miixe xed d P Prriiva vatte e -- P Pub ublliic c M Miix xe ed d P Pub ublic lic -- P Priv riva ate te Ratio of the mean Figure 3 intra-couplet interval to the grand mean for each couplet category by age group and couplet year Ratio of the mean intra-couplet interval to the grand mean for each couplet category by age group and couplet year. Table 3: Loss from each payment classification as a function of age and intra-couplet interval Rate of Loss (percent per intra-couplet year) Difference in Rate (percent per intra-couplet year) Public Private Insured Public vs. Private Private Insured Age group* All intervals* All intervals* 0 to 13 yrs 14 to 21 yrs All Intervals* 0 to 13 vs. 14 to 21 yrs 0–16 Years 0.63 1.77 2.37 0.40 1.14 1.97 17–39 Years 0.49 2.27 2.39 0.55 1.78 1.84 40–69 Years 0.81 1.94 2.56 0.46 1.13 2.10 70+ Years 0.25 2.64 2.67 1.91 2.39 0.76 * Averaged over all intra-couplet intervals in years. Maximum intra-couplet interval 16 years as age determined at second episode. exerting their preference. Cognitive explanations include ant upon marketed value propositions than use of Medi- the possibility that PHI may not be as entrenched cultur- care. In other words, PHI may be perceived by the ally as Medicare (the public system) in this population populace as a market good, whereas the public system and, as a consequence, use of PHI may be more depend- (Medicare) may be perceived as a fundamental right. Page 8 of 11 (page number not for citation purposes) 1981 1981 1981 1981 1981 1981 1981 1981 1983 1983 1983 1983 1983 1983 1983 1983 1985 1985 1985 1985 1985 1985 1985 1985 1987 1987 1987 1987 1987 1987 1987 1987 1989 1989 1989 1989 1989 1989 1989 1989 1991 1991 1991 1991 1991 1991 1991 1991 1993 1993 1993 1993 1993 1993 1993 1993 1995 1995 1995 1995 1995 1995 1995 1995 1997 1997 1997 1997 1997 1997 1997 1997 1999 1999 1999 1999 1999 1999 1999 1999 2001 2001 2001 2001 2001 2001 2001 2001 1981 1981 1981 1981 1981 1981 1983 1983 1983 1983 1983 1983 1985 1985 1985 1985 1985 1985 1987 1987 1987 1987 1987 1987 1989 1989 1989 1989 1989 1989 1991 1991 1991 1991 1991 1991 1993 1993 1993 1993 1993 1993 1995 1995 1995 1995 1995 1995 1997 1997 1997 1997 1997 1997 1999 1999 1999 1999 1999 1999 2001 2001 2001 2001 2001 2001 Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 0to16Years 17 to 39 Years 100 100% % 100 100% % 80 80% % 80 80% % 60 60% % 60 60% % 40 40% % 40 40% % 20 20% % 20 20% % 0% 0% 0% 0% 12 1# 2# 33# 44# 55# 66 77# 88# 99 1010 1111 11# 23 2 3# 44# 567 5#6#7# 89 8#9# 1100 1111 40 to 69 Years 70 + Years 100 100% % 100 100% % 80 80% % 80 80% % 60 60% % 60 60% % 40 40% % 40 40% % 20 20% % 20 20% % 0% 0% 0% 0% 12 1 2# 33# 44# 55# 66# 7#7 8#8 9#9 10#10 1111 12 1 2# 33# 44# 56 5# 6# 77# 88# 99# 110#0 1111 Intra-couplet interval (Yrs) Intra-couplet interval (Yrs) nd nd 2 Couplet Episode 1981 - 1990 2 Couplet Episode 1991 - 2001 Th Figure 4 e proportionate discordance among hospital couplets with a private first episode by decade of the second couplet episode The proportionate discordance among hospital couplets with a private first episode by decade of the second couplet episode. # Significant difference (p < 0.01) between the percentage of discordant second episodes occurring in 1981–1990 versus 1991– We also found that the degree of switching from PHI has been a major focus of recent federal government pol- towards the public system was inversly proportional to icy, with the introduction of a lifetime community rating the length of time between episodes. Possibly, healthier in 2000 in an effort to encourage younger, healthier indi- individuals were more likely to switch to the public viduals to take out and remain in PHI funds. Our finding system than sicker individuals, assuming that a relatively that in all age groups the overall rate of switching away short duration of intra-couplet interval can be taken as an from PHI was slightly higher in the period 1981 – 1990 indicator of increased morbidity. This phenomenon is compared with the rate in the period 1991 – 2001, sug- consistent with reports that the decline in the proportion gests that these policies have had an effect on behaviour of the eligible population holding PHI since the introduc- consistent with the government's intention. tion of Medicare in 1984 has been largely attributed to younger and healthier individuals dropping out [11,12]. An alternative, but less likely explanation for our findings, This finding indicated the presence of a substantial cross- may be systematic differences in the cognitive decision subsidisation between low risk and high risk individuals making process between long intra-couplet intervals com- further exacerbating the adverse selection price 'death spi- pared with between shorter intra-couplet intervals. For ral' of PHI [11]. Improving the risk profile of PHI holders example, individual historical preferences may play a role Page 9 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 in short interval switching but may not be important over not been reported previously in the literature and repre- longer intervals, where decisions may be made in isola- sents a new method of analysing longitudinal data on use tion. While it is important to consider this alternative, we of health insurance. The couplet methodology has ena- feel that if the changes observed in switches away from bled patterns to be measured based the behaviour of indi- PHI were largely due to differences in cognitive decision viduals rather than average shifts in private – public mix making, one would expect to see a similar phenomenon generated from unlinked episodes of care. However, we in switching away from a public classification. Such a phe- recognise that since the analysis was based on episodes of nomena was not observed in our data. care, patient initiated switching as part of a hospital trans- fer could not be analysed. This limitation may have In all age groups, our analysis indicated that the overall affected our conclusions about intra-couplet intervals. rate of switching from the private payment classification was slightly higher in the period 1981 – 1990 compared This paper is dedicated to explaining the couplet tech- with the rate in the period 1991 – 2001, suggesting that nique for the first time and applying it to address an initial the policies had an effect on behaviour consistent with the set of relatively descriptive questions (ie teasing out what government's intention. is happening). The next stage of investigation is a more analytic analysis which recognises that a wide range of Assumptions and limitations of the approach potential explanatory variables might predict the couplet- This study made use of the WA Data Linkage Project based phenomenon of switching (the why). For example which is unique in Australia and is one of only a small hospitalisation rates in set periods before and after a cou- number of population-based record linkage systems in plet, stratified by different lengths of stay and/or different the world. The use of administrative data has strengths DRG weights. In addition, the admission types (emer- and weaknesses. For example data can be inaccurate due gency/elective) of the members of the couplet may also be to recording or coding errors [17] or linkage errors. For important predictors. We feel that the couplet methodol- this study individual patient records were linked by prob- ogy described in this paper will enable significant inroads abilistic matching, using an automated computer algo- to be made into the investigation of such explanatory rithm based on the probability of two records being from variables. different people having the same identifier and two records from the same person having different identifiers. Conclusion The probabilities were then aggregated into a score and Our study found that the population of Western Australia checked against a threshold to determine if a match was exhibited distinct behavioural patterns in the switching of made. This technique typically has been found to have a payment classifications for inpatient hospitalisation true positive predictive value of 95–99% and a negative between 1980 and 2001. Private patients were more likely predictive value of 98–99% [18]. Extensive validation of to switch payment classification than public patients with the quality of the performance of matching has been shorter intervals between episodes corresponding to a undertaken on the WA Record Linkage Project using sam- greater probability of private-to-public switching. How- pling techniques and the proportions of mismatches and ever, the average rate of switching from a privately insured missed matches found were in the order of approximately classification was greater between 1981 and 1990 than 0.11% [18]. between 1991 and 2001, indicating that recent health care policy reforms implemented by the federal government to Linked data have the advantage of supporting a large and promote uptake of PHI have had an impact on behaviour. diverse research programme at relatively low cost, once the infrastructure is in place. They have the capacity to Competing interests provide a population-based view of events experienced Professor D'Arcy Holman is an independent director of longitudinally by individuals across all institutions [17]. HBF Health Funds inc which is the largest provider of pri- Given the objectives of this study, the latter point makes vate health insurance in Western Australia. the use of linked data particularly appropriate. Authors' contributions The approach we have taken in order to analyse the use of The manuscript has been read and approved by all PHI and Medicare by the population of Western Australia authors and the requirements for authorship have been is unique in at least two respects. Firstly the study was con- met as outlined below. REM was responsible for the ducted at the population level, due to the use of hospital conception and design of the study; analysis and interpre- morbidity data. Therefore, the "reference population" was tation of the data; and drafting and revising the paper. not merely an abstract concept as in conventional quanti- CDJH was responsible for conception and design of the tative research, but an operationalised descriptor of the study; interpretation of the data; and revising the paper. study sample. Secondly, our 'couplet methodology' has Page 10 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Acknowledgements The initial construction of the Data Linkage System was funded by the Western Australian Lotteries Commission. We would like to thank the WA Department of Health for on-going support of the Data Linkage Unit. References 1. Costa J, Garcia J: Demand for Private Health Insurance: How Important is the Quality Gap? Health Economics 2003, 12:587-599. 2. Cromwell D: The Lore about Private Health Insurance and Pressure on Public Hospitals. Australian Health Review 2002, 25(6):72-74. 3. Deeble J: The Private Health Insurance Rebate: Report to State and Territory Health Ministers. National Centre for Epi- demiology and Population Health The Australian National University; 4. McAuley IA: Stress on Public Hospitals – Why Private Insur- ance Has Made it Worse. University of Canberra: Discussion Paper: Australian Consumers' Association and the Australian Health- care Association; 2004. 5. Duckett SJ, Jackson TJ: The New health Insurance Rebate: An Inefficient Way of Assisting Public Hospitals. Medical Journal of Australia 2000, 172:439-442. 6. Willcox S: Promoting Private Health Insurance in Australia. Health Affairs 2001, 20(3):152-161. 7. Access Economics: Striking a Balance: Choice, Access and Affordability in Australian Health Care. APHA 2002. 8. Harper I: Preserving Choice: A Defence of Public Support for Private Health Care Funding in Australia. Medibank Private 9. Econotech Pty Ltd, Harper Associates, Hagan P: Easing the Pres- sure: The Intergenerational Report and Private Health Insurance. Medibank Private 2004. 10. Gross P: The Value Proposition for Private Health Insurance and the Private Health Sector in Australia: A Framework for Public Debate about Choices. St Christophe en Brionnais, Saone et Loire, France: Health Group Strategies Pty Limited & Institute of Health Economics and Technology Assessment; 2004. 11. Butler J: Policy Change and Private Health Insurance: Did the Cheapest Policy do the Trick? Australian Health Review 2002, 25(6):33-41. 12. Cormack M: Private Health Insurance: The Problem Child Faces Adulthood. Australian Health Review 2002, 25(2):38-51. 13. Holman CDJ, Bass AJ, Rouse IL, Hobbs MST: Western Australia: Development of a Health Services Research Linked Database. Aust NZ J Public Health 1999, 23(5):453-459. 14. Blewett N: The Politics of Health. Australian Health Review 2000, 23(2):10-19. 15. Duckett SJ: The Australian Health Care System 2nd edition. Oxford: Oxford University Press; 2004. 16. Moorin R, Holman CDJ: A longitudinal study of in-patient insur- ance classification in Western Australia using linked hospital morbidity data. Perth: The University of Western Australia; 2004. 17. Armstrong BK, Kricker A: Editorial: Record Linkage – A Vision Renewed. Australian and New Zealand Journal of Public Health 1999, 23(5):451-452. 18. Holman CDJ: The Analysis of Linked Health Data: Principles and Hands- On Applications. Dec 2002 edition. School of Population Health, Uni- versity of Western Australia; 2002. Publish with Bio Med Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical researc h in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 11 of 11 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australia and New Zealand Health Policy Springer Journals

Patient-initiated switching between private and public inpatient hospitalisation in Western Australia 1980 – 2001: An analysis using linked data

Loading next page...
 
/lp/springer-journals/patient-initiated-switching-between-private-and-public-inpatient-cgJBhcFpah
Publisher
Springer Journals
Copyright
Copyright © 2005 by Moorin and Holman; licensee BioMed Central Ltd.
Subject
Medicine & Public Health; Public Health; Social Policy
eISSN
1743-8462
DOI
10.1186/1743-8462-2-12
pmid
15978139
Publisher site
See Article on Publisher Site

Abstract

Background: The aim of the study was to identify any distinct behavioural patterns in switching between public and privately insured payment classifications between successive episodes of inpatient care within Western Australia between 1980 and 2001 using a novel 'couplet' method of analysing longitudinal data. Methods: The WA Data Linkage System was used to extract all hospital morbidity records from 1980 to 2001. For each individual, episodes of hospitalisation were paired into couplets, which were classified according to the sequential combination of public and privately insured episodes. Behavioural patterns were analysed using the mean intra-couplet interval and proportion of discordant couplets in each year. Results: Discordant couplets were consistently associated with the longest intra-couplet intervals (ratio to the average annual mean interval being 1.35), while the shortest intra-couplet intervals were associated with public concordant couplets (0.5). Overall, privately insured patients were more likely to switch payment classification at their next admission compared with public patients (the average rate of loss across all age groups being 0.55% and 2.16% respectively). The rate of loss from the privately insured payment classification was inversely associated with time between episodes (2.49% for intervals of 0 to 13 years and 0.83% for intervals of 14 to 21 years). In all age groups, the average rate of loss from the privately insured payment classification was greater between 1981 and 1990 compared with that between 1991 and 2001 (3.45% and 3.10% per year respectively). Conclusion: A small but statistically significant reduction in rate of switching away from PHI over the latter period of observation indicated that health care policies encouraging uptake of PHI implemented in the 1990s by the federal government had some of their intended impact on behaviour. Page 1 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 the Health Insurance Act of 1973. Thus switches between Background Coexistence of public and private health insurance, such use of the public and PHI system are initiated solely by as in Australia, has been the subject of intense debate patients based on choice rather than being mandated by among health economists and policy makers [1]. The government, notwithstanding that this 'choice' may be main issue surrounding this debate has been how the mix constrained in some instances by socioeconomic and of public and private health care financing influences the locational access factors. We propose that since posses- demand for private health insurance (PHI) and whether sion and utilisation of PHI are not equivalent, analysing PHI takes pressure off the public system [2]. the effectiveness of the recent government strategies in relieving the pressure on the public system cannot be Falling PHI membership, observed since the introduction accomplished by evaluations of changes in possession of of Medicare in 1984, was thought to have increased the PHI alone. Rather, changes in choice, as reflected by demand on the public system [3], prompting the federal patient-initiated switching between the public and private government to implement policies aimed at encouraging insurance systems must be analysed. possession of PHI to take the pressure off public hospitals and restore balance to the health care system [3-5]. Since The aim of this study was to identify and measure changes 1995 three major policy reforms have been introduced in in the behavioural patterns of switching between public Australia [6]. Firstly, in 1995 selective contracting was and privately insured status for hospitalisation by the introduced. The then federal Labor government passed population of Western Australia using our novel couplet legislation allowing private health plans to contract selec- methodology for analysing longitudinal data. tively with hospitals and doctors so as to improve compe- tition on price and quality. Secondly, government Method subsidisation of PHI was introduced in 1997 by the Con- Hospital morbidity data extraction and case selection servative coalition federal government as a means-tested The WA Data Linkage System [13] was used to extract all st rebate, capped at a flat amount irrespective of the cost of hospital morbidity data system (HMDS) records from 1 st PHI, combined with an income tax surcharge for high January 1980 to 31 December 2001, containing income earners without PHI. The rebate part of the policy encrypted patient identification and episode numbers, was subsequently replaced in 1999 by a non-means-tested age, gender, date of admission, date of separation, pay- 30% rebate for PHI available to everyone. Finally, in ment classification (public, insured private, or "other"), 2000, lifetime community rating was introduced. This and hospital type. The "other" payment category, which policy relaxed the previous stringent community rating included the private uninsured (2.2% of the total epi- system by allowing the price of PHI to be varied according sodes), workers compensation (1.8%), motor vehicle to the age at which a member joined [6]. (0.7%), defence force personnel (0.3%) and Veteran Affairs (1.7%) classifications was removed from the data The Australian Healthcare Agreement 1998–2003 com- set, leaving only the categories of public and private mitted the Commonwealth and states to review the rela- insured. This was done because the study was concerned tionship between PHI cover and the use of hospital with elective shifts between PHI and public categories; not services by private patients. The investigation of this rela- enforced payment classifications due to mandatory fund- tionship is becoming a priority as there is disagreement ing arrangements, or private episodes for which the among commentators as to the financial efficiency of the patients paid the full cost. 30% rebate [5,7-10]. To date, analyses of the effects of pol- icies aimed at supporting PHI in Australia have primarily Data coverage centred on changes in the proportion of the population This research has made use of records from the HMDS covered by PHI [3,6,11,12]. However, changing the pro- which is the inpatient information system for WA acute portion of the population covered may not directly trans- care hospitals. The data collected by the HMDS are patient late to increased utilisation and, therefore, reduce pressure identification, socio-demographic, services, administra- on the public system. The relationship between PHI cover tion and clinical diagnosis information. Every WA hospi- and type of hospital use is complex because the universal- tal defined as an acute care facility has to provide data via ity of Medicare in Australia means that anyone can be the Health Act or for private hospitals as part of their treated in a public hospital at no charge regardless of their license and every free-standing day surgery unit must also insurance status. provide data. The HMDS includes patient information for all acute A class hospitals, day surgery units, geriatric and The right of individuals to choose between public and pri- psychiatric inpatient facilities in general hospitals and, as vate insurance, regardless of the status (public or privately of July 1993, healthy newborn infants born in hospital. It insured) of previous hospitalisations or the possession of does not include patients of stand-alone geriatric and psy- PHI is protected by the principles of Medicare as set out in chiatric institutions, geriatric hostels or rehabilitation Page 2 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Couplet 1 Couplet 2 Episode of Hospitalisation Individual Patient Record 1st Episode 2nd Episode 3rd Episode Th Figure 1 e formation of hospital couplets The formation of hospital couplets. units; however, most metropolitan hospitals have geriat- Stratification by age group ric and psychiatric units attached for which data are pro- Stratification was based on age at admission of the first vided for the HMDS. Over the lifetime of this study there episode in each couplet. Each episode of care was assigned were no significant changes in the HMDS coverage. to one of four broad age categories (0–16 years, 17–39 years, 40–69 years and 70+ years) chosen to represent PHI Formation of episodes of care and assignment of payment market segmentation (children, young adults, middle age classification and old age) following consultation with a local private For each individual in the data set all eligible hospital health insurer. records were grouped into episodes of care, using the sep- Formation and classification of hospital couplets aration and admission dates to define temporally contig- uous periods of health care service utilisation. Thus one For each individual in the data set, eligible episodes of episode of care could have contained one or more inter care were grouped incrementally, starting with the index hospital transfers. Each episode of care was assigned to (first episode of care) to form hospital couplets such that one of the two eligible payment classifications (public or episodes 1 and 2 formed hospital couplet 1, episodes 2 privately insured) on the basis of the initial payment clas- and 3 formed hospital couplet 2 and so on (see figure 1). sification at admission, where inter-hospital transfers Hospital couplets were categorised depending upon the were involved. Allocation of public or PHI status was sequential combination of payment classifications of based solely on payment classification and not on hospi- their two contributing episodes of care as follows: tal type. This was done because the focus of this paper (consistent with the focus of public policy making in Aus- (i) Concordant public couplets containing only public tralia) was on use of PHI rather than hospital type. episodes of care. Page 3 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 (ii) Concordant private couplets containing only privately recorded for these individuals formed 6,376,472 distinct insured episodes of care. hospital couplets (see table 1). Of these, approximately one half (51%) had an intra-couplet interval of one year (iii) Discordant (mixed) public to private couplets had or greater. Details of the distribution of the hospital cou- the first episode of care as public and the second as pri- plets by intra-couplet interval are shown in table 2. Signif- vately insured. icant differences were observed in the characteristics of couplets having intra-couplet intervals of less than 1 year (iv) Discordant (mixed) private to public couplets had the compared with those of one year or over. The largest dif- first episode of care as privately insured and the second as ferences were observed in relation to hospital type and public. couplet category. Analysis of the mean intra-couplet interval The distribution of first couplet episodes in each year of The intra-couplet interval was defined as the time in days observation was more uniform in couplets with less than between the final separation from the first episode of care a one year intra-couplet interval compared with couplets and admission to the second episode of care. The mean longer intra-couplet interval or all couplets in the data file intra-couplet interval was calculated for each couplet clas- as shown in figure 2(A). The reduction of first couplet epi- sification by the year of admission of the second episode sodes in couplets with intervals = one year was directly of care (a term we subsequently refer to as the couplet proportional to the number of years remaining in which a year) and expressed as a ratio of the sum of the mean second episode (thus completing a couplet) could be intra-couplet interval of all couplets in that year, regard- observed. The lack of first episodes observed in 1980 was less of classification. We subsequently refer to this meas- the result of a reduced volume of data in the original ure as the grand mean. Thus a ratio greater than one for a HMDS file for that year, most likely caused by an extrac- particular couplet classification was indicative of an intra- tion error. couplet interval longer than the grand mean for that cou- plet year. The couplet year indicates the year the switch Figure 2(B) shows the distribution of second episodes in (choice) was made. couplets over the observation period. The proportion of second episodes increased over the observation period Analysis of behavioural patterns in switching of payment regardless of the duration of the intra-couplet interval. classification This was a function of the increased number of individu- The proportion of each type of discordant couplet relative als eligible to complete a couplet with a second episode as to the total number of couplets having a first episode of time progressed. care in the baseline payment classification was deter- mined independently for all intra-couplet intervals (in Mean intra-couplet interval whole years) in the data set. The ratio of the mean intra-couplet interval observed for each couplet category relative to the grand mean by age This analysis was performed separately for each age group group and couplet year is shown in figure 3. Discordant for the whole observation period and two predefined time couplets had the longest intra-couplet intervals, having on periods (1981 to 1989 and 1990 to 2001) chosen to rep- average a ratio relative to the annual grand mean intra resent the two main eras of health care policy in Australia. couplet interval of 1.35, while concordant couplets types The first related to the removal and re-introduction of free had the shortest intra-couplet intervals, their ratio being public hospital care, while the second related to changes 0.65. The overall pattern indicated that the longer the time in federal health policies aimed at supporting PHI between the first and second episode of a couplet, the [5,12,14,15]. Hospital couplets were partitioned into the more likelihood there was of a change in payment classi- two time periods using the year of admission of second fication, especially where the first payment classification episodes of care. The values obtained were plotted as seg- was private. The trends also indicated that, within each mented trend lines and the average rates of loss from each age group, individuals with public concordant couplets, payment classification per year of intra-couplet interval on average, had shorter intervals between episodes (ratio were calculated using least squares fit. 0.5) than individuals with private concordant couplets (ratio 0.8). Results Characteristics of the hospital couplets in the HMDS data The removal / re-introduction of free public hospital care file (1980 – 1984) has previously been shown to have had a The HMDS data file contained data pertaining to significant effect on use of in-patient insurance classifica- 1,979,946 individuals of which 1,185,014 (60%) had at tions [16]. In this study no significant difference was least one valid couplet. The 7,561,486 episodes of care observed with regard to when the first episode of a couplet Page 4 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Table 1: Characteristics of individuals, hospital episodes and hospital couplets nd Characteristic Individuals Episodes Couplets (2 Episode) Number % of Dataset Number % of Dataset Number % of Dataset Sex Male 522167 44.1 3156438 41.7 2634271 41.3 Female 662841 55.9 4405035 58.3 3742194 58.7 Indeterminate 4 0.0 13 0.0 7 0.0 TOTAL 1185014 100 7561486 100 6376472 100 Age Group 0–16 Years 313244 26.4 1045092 13.9 731848 11.5 17–39 Years 443602 37.4 2495039 33.1 2051437 32.3 40–69 Years 326974 27.6 2708572 35.9 2381598 37.5 70+ Years 100870 8.5 1292462 17.1 1191592 18.7 1 2 3 7541165 100 6356475 100 TOTAL 1184690 100 Hospital Teaching 367964 31.1 2852257 37.7 2484293 39.0 Type Public Metropolitan 198456 16.7 902286 11.9 703830 11.0 Private Metropolitan 313563 26.5 1838354 24.3 1524791 23.9 Public Country 279315 23.6 1745527 23.1 1466212 23.0 Private Country 25097 2.1 208883 2.8 183786 2.9 4 5 6 TOTAL 1184395 100 7547307 100 6362912 100 324 missing 20321 missing 19997 missing 619 missing 14179 missing 13560 missing Missing records were caused by missing data in the relevant fields of the individual records and do not reflect deficiencies in the linkage process. occurred (pre or post Medicare); however, the point in Behavioural patterns in switching of payment time of the second episode did correspond with behav- classification ioural change as can be seen in figure 3. Differences were observed in the rate of loss from the pub- lic and privately insured payment classifications across Differences in trend were observed across the four age age groups as shown in table 3. Across all age groups the groups. The 0–16 years age group trended towards a largest losses, at all intra-couplet intervals, occurred from reduction in the intra-couplet interval associated with dis- the privately insured payment classification. The largest cordant private to public couplets after 1985. This was not differences in the rates of loss were observed in the 70+ observed during the late 1980s and early 1990s in the years and the 17–39 years age groups, where an additional other three age groups. The 40–69 years age group showed 2.39 and 1.78 percent of privately insured episodes, a trend pattern similar to the 17–39 years age group until respectively, were lost for every year of intra-couplet inter- 2000, when the average intra-couplet interval associated val. The rates of loss from the privately insured payment with discordant private to public couplets reduced classification over shorter intra-couplet intervals (defined sharply. In addition there was a sharp increase in the aver- as 0 to 13 years) were greater than the rates of loss over age intra-couplet interval associated with discordant pub- longer intra-couplet intervals (defined as greater than 14 lic to private couplets (also observed in the 0 to 16 years years) in all age groups with the largest difference being age group) observed at that time. In the oldest age group observed in the 40–69 years age group and the smallest in (70+ years) the average difference in intra-couplet interval the 70+ years age group. The definition of short and long between public and private concordant couplets was intra-couplet interval was based on an observed substan- much smaller than observed in any of the other three age tial change of slope (inflection) in the segmented trend groups. lines. Over all age groups the trend in concordant couplets was Figure 4 shows the degree of switching from private to that of a slowly reducing average intra-couplet interval, public episodes over the two designated eras in health care excluding the 1983 – 1984 period, where the average policy. In all age groups the average rate of switching away intra-couplet interval for private concordant couplets from the private sector in 1991–2001 (era 2) was lower increased in all age groups. than that observed in 1981–1990 (era 1). The decrease in Page 5 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Table 2: Distribution of hospital couplets by characteristics and intra-couplet interval Characteristic Intra-Couplet Interval (I-CI) Less than 1 Year 1 Year or greater Number % in Dataset % Across I-CI Number % in Dataset % Across I-CI Sex Male 1397957 44.6 53.1* 1236314 38.2 46.9 Female 1739643 55.4 46.5* 2002551 61.8 53.5 Indeterminate 5 0.0 71.4 2 0.0 28.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 Age Group 0–16 Years 318255 10.2 43.5* 413593 12.8 56.5 17–39 Years 871234 27.9 42.5* 1180203 36.5 57.5 40–69 Years 1271495 40.7 53.4* 1110103 34.4 46.6 70+ Years 664439 21.3 55.8* 527153 16.3 44.2 1 2 TOTAL 3125423 100 49.2* 3231052 100 50.8 Hospital Type Teaching 1554708 49.6 62.6* 929585 28.7 37.4 Public Metropolitan 249979 8.0 35.5* 453851 14.0 64.5 Private Metropolitan 573729 18.3 37.6* 951062 29.4 62.4 Public Country 672844 21.4 45.9* 793368 24.5 54.1 Private Country 77205 2.5 42.0* 106581 3.3 58.0 Other 9140 0.3 67.4* 4420 0.1 32.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 Couplet Category Concordant Public 2104333 67.1 56.7* 1605368 49.6 43.3 Private 887622 28.3 43.0* 1176629 36.3 57.0 Discordant Public – Private 74132 2.4 31.5* 161140 5.0 68.5 Private – Public 71518 2.3 19.4* 296730 9.1 80.6 TOTAL 3137605 100 49.2* 3238867 100 50.8 * Statistically significantly different (at the 0.05% level) to the proportion of couplets with an intra-couplet interval 1 year or greater. 12182 missing 7815 missing rate was small (average across all age groups -0.35% per in their next admission than public patients, irrespective intra-couplet year) but statistically significant. In addi- of the length of time between the two episodes. There are tion, significance testing of the difference between each a number of possible explanations for this including puta- pair of proportions (era 1 versus era 2) indicated a signif- tive structural and cognitive reasons for the observed icant difference for the majority as indicated in figure 4. behaviour. Structurally, the average patient who begins with a public classification is likely to be of lesser socioe- We also found that as the intra-couplet interval increased, conomic means than the average patient who begins with the difference between the proportions of discordant cou- a private insured classification. The option of the former plets having a private first episode versus a public first epi- patient to use PHI at the next hospitalisation will be sode increased from 4.2% at one year to 34.3% at 18 years dependant upon them taking out, or at the very least (data not shown). This indicated that overall, regardless of maintaining (if they had private cover at the initial epi- age, there was greater switching away from PHI than away sode but did not use it) private cover in the meantime. The from the public classification. same pre-requisite does not exist for an initially privately insured patient accessing a public classification on the next occasion. Also patients with private insurance experi- Discussion As expected, due to their greater capacity to move between encing trauma or an acute disease event may, in some cir- respective payment classifications, private patients were cumstances, be admitted in an emergency as a public found to be more likely to switch payment classifications patient, thus to some extent disallowing the patient from Page 6 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 14 14 14 14% % % % 12 12 12 12% % % % 10 10 10 10% % % % 8% 8% 8% 8% st st %o %off 11 Ep Epiis sode odes s 6% 6% 6% 6% in in Co Cou up plet lets s 4% 4% 4% 4% 2% 2% 2% 2% 0% 0% 0% 0% st st Y Year ear o off 1 1 C Coup ouplle ett E Epi pis so ode de 10 10 10 10 10 10 10 10% % % % % % % % Al Al Al Al Al Al Allllllll C C C C C C Co o o o o o ou u u u u u up p p p p p pllllllle e e e e e et t t t t t ts s s s s s s > >= = 1 1 1 1 1 Y Y Y Y Ye e e e ea a a a ar r r r r IIIIInt nt nt nt ntr r r r ra a a a a - - - - ---C C C C C C Cou ou ou ou ou o ou up p p p p p pllllllle e e e e et et t t t t t IIIIIIInt nt nt nt nt n nte te e e e e er r r r r r rv v v v v v val al a a a a alllll 8% 8% 8% 8% 8% 8% 8% 8% < < < < < < < 1 1 1 1 1 1 1 Y Y Y Y Y Y Ye e e e e e ea a a a a a ar r r r r r r In In In In In In Intttttttr r r r r r ra a a a a a a -------C C C C C C Co o o o o o ou u u u u u up p p p p p ple le le le le le lettttttt IIIIIIIn n n n n n nttttttte e e e e e er r r r r r rv v v v v v va a a a a a alllllll %o %o %o %o %offfff 2 2 2 2 2n n n n nd d d d d 6% 6% 6% 6% 6% 6% 6% 6% Epi Epi Epi Epi Epis s s s sod od od od ode e e e es s s s s in in in in in C C C C Co o o o ou u u u up p p p pllllle e e e ets ts ts ts ts 4% 4% 4% 4% 4% 4% 4% 4% 2% 2% 2% 2% 2% 2% 2% 2% 0% 0% 0% 0% 0% 0% 0% 0% Distribution Figure 2 of the proportion of first and second episodes in couplets by year Distribution of the proportion of first and second episodes in couplets by year. Page 7 of 11 (page number not for citation purposes) 198 198 1980 0 0 19 19 19 19 19 19 19 198 8 8 8 8 8 8 80 0 0 0 0 0 0 0 19 19 1982 82 82 19 19 19 19 19 19 19 198 8 8 8 8 8 8 82 2 2 2 2 2 2 2 19 19 1984 84 84 19 19 19 19 19 19 19 198 8 8 8 8 8 8 84 4 4 4 4 4 4 4 19 19 1986 86 86 19 19 19 19 19 19 19 198 8 8 8 8 8 8 86 6 6 6 6 6 6 6 1 1 1988 988 988 19 19 19 19 19 19 19 198 8 8 8 8 8 8 88 8 8 8 8 8 8 8 1 1 1990 990 990 19 19 19 19 19 19 19 199 9 9 9 9 9 9 90 0 0 0 0 0 0 0 1 1 1992 992 992 19 19 19 19 19 19 19 199 9 9 9 9 9 9 92 2 2 2 2 2 2 2 1 1 1994 994 994 1 1 1 1 1 1 1 1994 994 994 994 994 994 994 994 19 19 1996 96 96 1 1 1 1 1 1 1 1996 996 996 996 996 996 996 996 1 1 1998 998 998 19 19 19 19 19 19 19 199 9 9 9 9 9 9 98 8 8 8 8 8 8 8 200 200 2000 0 0 20 20 20 20 20 20 20 200 0 0 0 0 0 0 00 0 0 0 0 0 0 0 Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 0- 0- 0-16 16 16 Y Y Year ear ears s s 17 17--3 39 9 Y Yea earrs s 2.0 2.0 2.0 2. 2. 2. 2.0 0 0 0 1.6 1.6 1.6 1. 1. 1. 1.6 6 6 6 1.2 1.2 1.2 1. 1. 1. 1.2 2 2 2 Ra Ra Ratttiiio o o = = = 1 1 1 Ra Ratio tio = = 1 1 0.8 0.8 0.8 0. 0. 0. 0.8 8 8 8 0.4 0.4 0.4 0. 0. 0. 0.4 4 4 4 0 0 0 0 0 0 0 4 4 40 0 0-6 -6 -69 9 9 Y Y Ye e ea a ars rs rs 7 70+ 0+ Y Year ears s 2. 2. 2. 2.0 0 0 0 2.0 2.0 2.0 1. 1. 1. 1.6 6 6 6 1.6 1.6 1.6 1. 1. 1. 1.2 2 2 2 1.2 1.2 1.2 Ra Ra Ratttiiio o o = = = 1 1 1 Ra Ratio tio = = 1 1 0. 0. 0. 0.8 8 8 8 0.8 0.8 0.8 0. 0. 0. 0.4 4 4 4 0.4 0.4 0.4 0 0 0 0 0 0 0 Publ Publiic c C Conc oncord ordant ant P Prriiv vat ate e C Conc onco orrd dant ant M Miixe xed d P Prriiva vatte e -- P Pub ublliic c M Miix xe ed d P Pub ublic lic -- P Priv riva ate te Ratio of the mean Figure 3 intra-couplet interval to the grand mean for each couplet category by age group and couplet year Ratio of the mean intra-couplet interval to the grand mean for each couplet category by age group and couplet year. Table 3: Loss from each payment classification as a function of age and intra-couplet interval Rate of Loss (percent per intra-couplet year) Difference in Rate (percent per intra-couplet year) Public Private Insured Public vs. Private Private Insured Age group* All intervals* All intervals* 0 to 13 yrs 14 to 21 yrs All Intervals* 0 to 13 vs. 14 to 21 yrs 0–16 Years 0.63 1.77 2.37 0.40 1.14 1.97 17–39 Years 0.49 2.27 2.39 0.55 1.78 1.84 40–69 Years 0.81 1.94 2.56 0.46 1.13 2.10 70+ Years 0.25 2.64 2.67 1.91 2.39 0.76 * Averaged over all intra-couplet intervals in years. Maximum intra-couplet interval 16 years as age determined at second episode. exerting their preference. Cognitive explanations include ant upon marketed value propositions than use of Medi- the possibility that PHI may not be as entrenched cultur- care. In other words, PHI may be perceived by the ally as Medicare (the public system) in this population populace as a market good, whereas the public system and, as a consequence, use of PHI may be more depend- (Medicare) may be perceived as a fundamental right. Page 8 of 11 (page number not for citation purposes) 1981 1981 1981 1981 1981 1981 1981 1981 1983 1983 1983 1983 1983 1983 1983 1983 1985 1985 1985 1985 1985 1985 1985 1985 1987 1987 1987 1987 1987 1987 1987 1987 1989 1989 1989 1989 1989 1989 1989 1989 1991 1991 1991 1991 1991 1991 1991 1991 1993 1993 1993 1993 1993 1993 1993 1993 1995 1995 1995 1995 1995 1995 1995 1995 1997 1997 1997 1997 1997 1997 1997 1997 1999 1999 1999 1999 1999 1999 1999 1999 2001 2001 2001 2001 2001 2001 2001 2001 1981 1981 1981 1981 1981 1981 1983 1983 1983 1983 1983 1983 1985 1985 1985 1985 1985 1985 1987 1987 1987 1987 1987 1987 1989 1989 1989 1989 1989 1989 1991 1991 1991 1991 1991 1991 1993 1993 1993 1993 1993 1993 1995 1995 1995 1995 1995 1995 1997 1997 1997 1997 1997 1997 1999 1999 1999 1999 1999 1999 2001 2001 2001 2001 2001 2001 Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 0to16Years 17 to 39 Years 100 100% % 100 100% % 80 80% % 80 80% % 60 60% % 60 60% % 40 40% % 40 40% % 20 20% % 20 20% % 0% 0% 0% 0% 12 1# 2# 33# 44# 55# 66 77# 88# 99 1010 1111 11# 23 2 3# 44# 567 5#6#7# 89 8#9# 1100 1111 40 to 69 Years 70 + Years 100 100% % 100 100% % 80 80% % 80 80% % 60 60% % 60 60% % 40 40% % 40 40% % 20 20% % 20 20% % 0% 0% 0% 0% 12 1 2# 33# 44# 55# 66# 7#7 8#8 9#9 10#10 1111 12 1 2# 33# 44# 56 5# 6# 77# 88# 99# 110#0 1111 Intra-couplet interval (Yrs) Intra-couplet interval (Yrs) nd nd 2 Couplet Episode 1981 - 1990 2 Couplet Episode 1991 - 2001 Th Figure 4 e proportionate discordance among hospital couplets with a private first episode by decade of the second couplet episode The proportionate discordance among hospital couplets with a private first episode by decade of the second couplet episode. # Significant difference (p < 0.01) between the percentage of discordant second episodes occurring in 1981–1990 versus 1991– We also found that the degree of switching from PHI has been a major focus of recent federal government pol- towards the public system was inversly proportional to icy, with the introduction of a lifetime community rating the length of time between episodes. Possibly, healthier in 2000 in an effort to encourage younger, healthier indi- individuals were more likely to switch to the public viduals to take out and remain in PHI funds. Our finding system than sicker individuals, assuming that a relatively that in all age groups the overall rate of switching away short duration of intra-couplet interval can be taken as an from PHI was slightly higher in the period 1981 – 1990 indicator of increased morbidity. This phenomenon is compared with the rate in the period 1991 – 2001, sug- consistent with reports that the decline in the proportion gests that these policies have had an effect on behaviour of the eligible population holding PHI since the introduc- consistent with the government's intention. tion of Medicare in 1984 has been largely attributed to younger and healthier individuals dropping out [11,12]. An alternative, but less likely explanation for our findings, This finding indicated the presence of a substantial cross- may be systematic differences in the cognitive decision subsidisation between low risk and high risk individuals making process between long intra-couplet intervals com- further exacerbating the adverse selection price 'death spi- pared with between shorter intra-couplet intervals. For ral' of PHI [11]. Improving the risk profile of PHI holders example, individual historical preferences may play a role Page 9 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 in short interval switching but may not be important over not been reported previously in the literature and repre- longer intervals, where decisions may be made in isola- sents a new method of analysing longitudinal data on use tion. While it is important to consider this alternative, we of health insurance. The couplet methodology has ena- feel that if the changes observed in switches away from bled patterns to be measured based the behaviour of indi- PHI were largely due to differences in cognitive decision viduals rather than average shifts in private – public mix making, one would expect to see a similar phenomenon generated from unlinked episodes of care. However, we in switching away from a public classification. Such a phe- recognise that since the analysis was based on episodes of nomena was not observed in our data. care, patient initiated switching as part of a hospital trans- fer could not be analysed. This limitation may have In all age groups, our analysis indicated that the overall affected our conclusions about intra-couplet intervals. rate of switching from the private payment classification was slightly higher in the period 1981 – 1990 compared This paper is dedicated to explaining the couplet tech- with the rate in the period 1991 – 2001, suggesting that nique for the first time and applying it to address an initial the policies had an effect on behaviour consistent with the set of relatively descriptive questions (ie teasing out what government's intention. is happening). The next stage of investigation is a more analytic analysis which recognises that a wide range of Assumptions and limitations of the approach potential explanatory variables might predict the couplet- This study made use of the WA Data Linkage Project based phenomenon of switching (the why). For example which is unique in Australia and is one of only a small hospitalisation rates in set periods before and after a cou- number of population-based record linkage systems in plet, stratified by different lengths of stay and/or different the world. The use of administrative data has strengths DRG weights. In addition, the admission types (emer- and weaknesses. For example data can be inaccurate due gency/elective) of the members of the couplet may also be to recording or coding errors [17] or linkage errors. For important predictors. We feel that the couplet methodol- this study individual patient records were linked by prob- ogy described in this paper will enable significant inroads abilistic matching, using an automated computer algo- to be made into the investigation of such explanatory rithm based on the probability of two records being from variables. different people having the same identifier and two records from the same person having different identifiers. Conclusion The probabilities were then aggregated into a score and Our study found that the population of Western Australia checked against a threshold to determine if a match was exhibited distinct behavioural patterns in the switching of made. This technique typically has been found to have a payment classifications for inpatient hospitalisation true positive predictive value of 95–99% and a negative between 1980 and 2001. Private patients were more likely predictive value of 98–99% [18]. Extensive validation of to switch payment classification than public patients with the quality of the performance of matching has been shorter intervals between episodes corresponding to a undertaken on the WA Record Linkage Project using sam- greater probability of private-to-public switching. How- pling techniques and the proportions of mismatches and ever, the average rate of switching from a privately insured missed matches found were in the order of approximately classification was greater between 1981 and 1990 than 0.11% [18]. between 1991 and 2001, indicating that recent health care policy reforms implemented by the federal government to Linked data have the advantage of supporting a large and promote uptake of PHI have had an impact on behaviour. diverse research programme at relatively low cost, once the infrastructure is in place. They have the capacity to Competing interests provide a population-based view of events experienced Professor D'Arcy Holman is an independent director of longitudinally by individuals across all institutions [17]. HBF Health Funds inc which is the largest provider of pri- Given the objectives of this study, the latter point makes vate health insurance in Western Australia. the use of linked data particularly appropriate. Authors' contributions The approach we have taken in order to analyse the use of The manuscript has been read and approved by all PHI and Medicare by the population of Western Australia authors and the requirements for authorship have been is unique in at least two respects. Firstly the study was con- met as outlined below. REM was responsible for the ducted at the population level, due to the use of hospital conception and design of the study; analysis and interpre- morbidity data. Therefore, the "reference population" was tation of the data; and drafting and revising the paper. not merely an abstract concept as in conventional quanti- CDJH was responsible for conception and design of the tative research, but an operationalised descriptor of the study; interpretation of the data; and revising the paper. study sample. Secondly, our 'couplet methodology' has Page 10 of 11 (page number not for citation purposes) Australia and New Zealand Health Policy 2005, 2:12 http://www.anzhealthpolicy.com/content/2/1/12 Acknowledgements The initial construction of the Data Linkage System was funded by the Western Australian Lotteries Commission. We would like to thank the WA Department of Health for on-going support of the Data Linkage Unit. References 1. Costa J, Garcia J: Demand for Private Health Insurance: How Important is the Quality Gap? Health Economics 2003, 12:587-599. 2. Cromwell D: The Lore about Private Health Insurance and Pressure on Public Hospitals. Australian Health Review 2002, 25(6):72-74. 3. Deeble J: The Private Health Insurance Rebate: Report to State and Territory Health Ministers. National Centre for Epi- demiology and Population Health The Australian National University; 4. McAuley IA: Stress on Public Hospitals – Why Private Insur- ance Has Made it Worse. University of Canberra: Discussion Paper: Australian Consumers' Association and the Australian Health- care Association; 2004. 5. Duckett SJ, Jackson TJ: The New health Insurance Rebate: An Inefficient Way of Assisting Public Hospitals. Medical Journal of Australia 2000, 172:439-442. 6. Willcox S: Promoting Private Health Insurance in Australia. Health Affairs 2001, 20(3):152-161. 7. Access Economics: Striking a Balance: Choice, Access and Affordability in Australian Health Care. APHA 2002. 8. Harper I: Preserving Choice: A Defence of Public Support for Private Health Care Funding in Australia. Medibank Private 9. Econotech Pty Ltd, Harper Associates, Hagan P: Easing the Pres- sure: The Intergenerational Report and Private Health Insurance. Medibank Private 2004. 10. Gross P: The Value Proposition for Private Health Insurance and the Private Health Sector in Australia: A Framework for Public Debate about Choices. St Christophe en Brionnais, Saone et Loire, France: Health Group Strategies Pty Limited & Institute of Health Economics and Technology Assessment; 2004. 11. Butler J: Policy Change and Private Health Insurance: Did the Cheapest Policy do the Trick? Australian Health Review 2002, 25(6):33-41. 12. Cormack M: Private Health Insurance: The Problem Child Faces Adulthood. Australian Health Review 2002, 25(2):38-51. 13. Holman CDJ, Bass AJ, Rouse IL, Hobbs MST: Western Australia: Development of a Health Services Research Linked Database. Aust NZ J Public Health 1999, 23(5):453-459. 14. Blewett N: The Politics of Health. Australian Health Review 2000, 23(2):10-19. 15. Duckett SJ: The Australian Health Care System 2nd edition. Oxford: Oxford University Press; 2004. 16. Moorin R, Holman CDJ: A longitudinal study of in-patient insur- ance classification in Western Australia using linked hospital morbidity data. Perth: The University of Western Australia; 2004. 17. Armstrong BK, Kricker A: Editorial: Record Linkage – A Vision Renewed. Australian and New Zealand Journal of Public Health 1999, 23(5):451-452. 18. Holman CDJ: The Analysis of Linked Health Data: Principles and Hands- On Applications. Dec 2002 edition. School of Population Health, Uni- versity of Western Australia; 2002. Publish with Bio Med Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical researc h in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 11 of 11 (page number not for citation purposes)

Journal

Australia and New Zealand Health PolicySpringer Journals

Published: Jun 27, 2005

References