  |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
02-May-2003, Vol 116 No 1173
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The cost to the New Zealand Government of providing
‘free’ primary medical care: an estimate based upon the Rand Health
Insurance Experiment
Tom Robinson
Accessible primary care is acknowledged as important for any
health system. Access is determined by a number of factors, including the cost
to the patient of using primary care
services.1,2 New Zealand has high user charges
for general practice services by the standards of many OECD countries. A number
of studies suggest that cost is a significant barrier to at least some New
Zealanders’ access to
GPs.3–7
The Primary Health Care Strategy makes it clear that the
funding of primary healthcare is on the Government’s
agenda.8 There are a number of important
advocates for free primary care. For example, in 2000 the National Health
Committee recommended that ‘The Government should preferentially invest in
primary health care services with the intention of moving to fully-funded care
over the next five years.’9
The cost to the Government of following such a policy is
unknown. This paper attempts to estimate this cost using available data on the
cost of health services and estimates of increased utilisation of health
services that are derived from the Rand Health Insurance Experiment (Rand
HIE).10–12 The Rand HIE remains the only
randomized controlled trial of the effect of user charges on health services
utilisation. Whilst the costs derived can be considered only as tentative
estimates, they are based upon the best information available and may lead to
useful discussion.
MethodsThis section necessarily
summarises the methods used to calculate the cost of free care. Those who would
like a more detailed account should contact the author.
The costs to Government of fully funding general practice care are considered under three categories:
The
costs in this article are calculated using 1998/99 year data. These were the
most recent data available to the author. Data on health service expenditure are
taken from the Ministry of Health publication
Health expenditure trends in New Zealand
1980–1999.13
Payment for current GP services It is assumed that in moving to free general practice care the Government would agree to pay general practices the same amount that the practices currently claim from patients and other private sources. Unfortunately, this amount is not known directly. Ministry of Health estimates are based upon data extrapolated from the Household Economic Survey.13 The cost of paying for new general practice services The cost of extra GP services used under a free system is dependent upon how much demand increases with a fall in price (the price elasticity of demand, or η). The best information that is available comes from the Rand HIE.10–12 Figure 1 shows the number of doctor visits demanded when care is subsidised as a ratio of the number demanded when patients pay the full cost. People who received free doctor visits utilised services just over 50% more than those who paid for all their care.  Figure 1. Demand for doctor visits with varying user
charges (from Rand HIE)
In New Zealand, many people do not pay the full cost of
GP visits (for example because they receive GMS or ACC subsidies, disability
allowance or have medical insurance). To estimate the increase in demand for
these groups should care become free, it is necessary to estimate their current
position on the graph, ie, what proportion of the total cost they currently pay
and thus what their current level of demand is as a ratio of full-cost demand.
By using this as a baseline we can estimate what their likely increase in
utilisation would be if they were to receive free care. For example, if a group
is currently paying half of the full-cost fee, their utilisation is estimated to
represent 117% of full-cost utilisation. If they were to receive free care, we
could expect them to represent 151% of full-cost utilisation. The increase in
demand for this group is therefore estimated to be 151/117, or 129% of their
current utilisation. In other words, we would expect their demand to increase by
29%.
The Rand HIE also clearly showed that although Figure 1 represents the overall population’s responsiveness to price changes, some groups are more or less responsive than this average.10,12 In particular, children and lower-income earners were more responsive to price changes than adults and higher-income earners. Therefore, in calculating the expected changes in utilisation, appropriate adjustments were made for these groups. Community services card (CSC) status was used as a proxy for income group. The Ministry of Health provided estimates of current consultation numbers. It was assumed that children under six years old already receive free care and therefore no change in demand is expected. This may not be entirely accurate, but a 2001 survey of 180 GPs by IMS Health showed that 70% of GPs never charge for children under six and only 5% always charge.15 Costs are otherwise calculated assuming a standard GP consultation charge of $38. This figure was estimated using a number of sources.16,17 Doctors’ responses to increased demand The Rand HIE determined utilisation changes due to a ‘pure demand’ effect of changes in user charges. This means it did not take into account doctors’ responses to the increases in demand. The patients enrolled in the experiment made up only a very small part of each doctor’s practice. Therefore, even if the enrolled patients increased their demand by 50%, their doctor would not notice any appreciable increase in workload and would be unlikely to respond. If, as predicted, the public’s demand for general practice consultations increases by 28% it seems likely that GPs would act to control this increased workload. Unfortunately, there is limited empirical evidence about this type of situation. Observational studies suggest that where changes in user charges have been introduced to an entire system (in Great Britain in 1948, Saskatchewan in 1968, and Montreal in 1970), changes in utilisation do seem to have been lower than the Rand HIE suggests.18–20 New Zealand studies of subsidy changes also suggest lower elasticities of demand.21,22 Because of these uncertainties, estimates are given for costs at two lower estimates of increased overall utilisation (14% and 7%, Scenarios B and C respectively). The price of complements A further complication of estimating changes in utilisation of general practice care is that demand is likely to be affected by the price of what economists call ‘complements’ as well as the price of GP consultations. A ‘complement’ in this context is any service or good that people usually purchase at the same time as the GP consultation. Important complements are medicines and diagnostic tests. The implication of this is that even if GP consultations are free, people will still perceive a cost in seeing a doctor if they expect to pay for a prescription or some other service. For example, if an adult without a CSC now expects to pay $38 for a consultation and $15 for a prescription (total $53), then under free primary care they will still perceive a cost of $15. If we make the assumption that patients base their demand on the price of the GP consultation and the cost of one prescription, we can recalculate the increase in demand for different groups using the same techniques as previously (Scenarios D and E). The cost of other health service utilisation changes The Rand HIE suggested that changes in demand for medicines and diagnostic tests were in direct proportion to changes in demand for doctor visits.12,14 It was therefore assumed that a 10% increase in the utilisation of GP consultations will lead to a 10% increase in the use of medicines and other referred services. The Rand HIE also found that decreasing user charges for doctors’ visits led to an increase in hospital admissions, although this increase was only about one fifth of the rise in visit utilisation (when hospital services were free throughout).10,11 No attempt was made to calculate what changes would occur in the use of mental health or disability support services as there is no information available on how the utilisation of these services alter with changing primary care utilisation. ResultsPayment
for current GP services In the year ending June 1999, $262 million was
spent by private sources on GP services ($197 million out-of-pocket and $65
million by insurance companies).13
The cost of paying for new
general practice services Based upon the Rand HIE, Table 1 shows that if
all user charges for general practice consultations were removed we could expect
an increase in the demand for consultations of between 19% and 47% depending
upon the group involved, and 28% overall. The cost of paying for these extra
consultations is calculated at $202 million.
Table 1. Calculation of increased number of
consultations and costs associated with free general practice
The cost of other health
service utilisation changes Table 2 shows costs for various referred
services for 1998/99 and calculated new costs given increased demand due to free
general practice care in the base scenario. Twenty eight per cent increase in
demand is used for public and private expenditure and 23% for ACC
expenditure.
Table 3 shows the equivalent costs for increased use of
medical and surgical hospital services. The cost of providing for the increased
utilisation of hospital services is calculated at $136 million.
Table 2. Referred service costs for free general
practice care, 1999 ($ millions)13
Table 3. Existing and new costs for institutional
medical and surgical care, 1999 ($
millions)13
Overall cost of free
general practice care The total cost of free general practice care to the
Government according to these calculations is in the order of $922 million
(Scenario A, Table 4).
Table 4. Total new costs to the government of free
general practice care (‘pure demand’ effect)
Doctors’ responses to
increased demand The above costs do not take into account doctors’
responses to increased demand. As discussed above, it is assumed that GPs do act
effectively to reduce this increase in utilisation by 50% or 75%. Therefore,
there is a 14% or 7% increase in use of GP services rather than 28%. The total
cost to the government with this scenario is therefore $592 or $428 million
(Scenarios B and C, Table 5).
Table 5. Total new costs to the government of free
primary care (with GP response)
The price of
complements Table 6 is calculated using the additional assumption that
patients base their demand on the price of the GP consultation and the cost of
one prescription. The overall increase in demand for GP consultations with this
assumption is 15% (compared with 28% previously). Again, this increase can be
halved to account for doctor’s response to this demand increase. Table 6
indicates that the cost to Government with these assumptions is $435 million or
$349 million (scenarios D and E).
Table 6. Total new costs to the Government of free
primary care (with GP response and accounting for continued prescription
costs
DiscussionBefore any major change in health
policy is made, it is important to at least estimate the likely costs and
benefits. This study has attempted to estimate the costs of providing free
primary care in New Zealand. Perhaps one of the outstanding results of
attempting this process is the realisation of the degree of uncertainty there
must be around these estimates. They are based upon limited data, estimates of
price elasticity that may not be valid in the New Zealand situation, and several
important assumptions.
Accounting for current user charges might seem to be the
simplest part of this exercise. Yet user charges for GP consultations are not
currently recorded on a national basis. The estimate used here is taken from the
Ministry of Health publication Health
Expenditure Trends in New Zealand 1980–1999 and this in turn is
extrapolated from the Household Economic
Survey.13 An alternative approach is to
multiply the estimated annual number of GP consultations by an estimate of user
charges in each group. User charges can be estimated by deducting the GMS
subsidy from the estimated average charge of $38. This approach, however, gives
a much greater figure of $433 million as the total cost of GP user charges. The
majority of the $171 million difference can probably be explained by the fact
that GPs often do not charge for consultations, or charge substantially less
than their normal fee. Tilyard and Dovey showed that in one region in 1993 GPs
charged their normal or advertised fee in only 37% of
cases.23
Estimates of increased GP use and referred service and
hospital use are based upon the Rand HIE. This was a randomized controlled trial
that compared health service utilisation among groups of people who were subject
to different co-payments. This paper assumes that the differences in utilisation
by groups facing different co-payments can reasonably be equated to changes in
demand that would occur if people faced equivalent co-payment changes. The Rand
HIE was carried out in several centres in the USA in the late 1970s. The elderly
and many people with disabilities were excluded from the study because they were
already covered by public health insurance. Using a 25-year-old study, performed
in the USA, which excluded important groups of the population is clearly less
than ideal. However, it is the only experimental study of this sort ever
performed and the only study that has looked at utilisation over a range of
user charges.
It is not surprising that decreasing the cost of general
practice consultations will lead to an increase in demand and utilisation of
these services. However, the shape of the curve in Figure 1 is perhaps less
predictable than one would think and has important implications. It shows that
the majority of patients’ response to user charges occurs with small
charges. The size of demand increases when shifting from full user charges to
25% user charges were much smaller than those that occurred when shifting from
25% user charges to free care. If this were to be repeated in New Zealand it
would have two important implications. First, even though many patient groups in
New Zealand already receive substantial subsidies for general practice care, if
their care became free we should expect large demand increases. Second, if the
Government did wish to limit demand increases, whilst reducing the financial
burden of primary medical care on households, this could be achieved by
maintaining modest user charges.
Whilst the Rand HIE provides the best available information
on increased demand for doctors’ consultations with decreasing user
charges, it tells us nothing about how doctors might respond to this increased
demand. It seems clear that doctors have some degree of control over utilisation
of their services, even when they have no control over
price.24–26 If changes in user charges
are introduced to an entire health system the size of utilisation changes may
well be smaller than expected.
When free care was introduced to Great Britain in 1948 it
was estimated that utilisation of physician visits, for those groups not
previously covered by National Health Insurance, increased between 10.7% and
16.2%.18 The introduction of free care in
Montreal in 1970 led to a 21.8% increase in utilisation of office or home
visits, balanced by a reduction in phone consultations and hospital-based care
(which was already free).19 The introduction of
a 33% co-payment for office visits in Saskatchewan in 1968 saw a fall in
utilisation that was estimated to be between 5.6% and
7.2%.20
In New Zealand, an opportunity to study the effect of moving
from subsidised GP consultations to free care for children under six years
occurred with the introduction of the Free Child Health Care Scheme (FCHCS) in
1997. An evaluation compared GP utilisation for the year prior to the
scheme’s introduction with the year following
introduction.21 Depending upon the database
analysed, the increase in utilisation was estimated at between 6% and 23%.
Another New Zealand study examined utilisation changes that occurred with
changes in GP consultation subsidies in 1992 and again suggested a lower price
elasticity of demand than the Rand HIE (about
18%).22 The degree to which doctors may wish to
exercise this control with the introduction of free care will depend upon the
circumstances in which it is introduced. If doctors are paid under a capitation
arrangement they will have a greater incentive to control demand than if they
are paid by fee-for-service. The supply, morale and workload of doctors will
have a strong impact upon the increased provision of services. The number of
doctors in primary care is currently declining, and if this were to continue it
would limit any increase in utilisation. Nurse and other primary care
professional consultations may be substituted for GP consultations and the
ability and willingness of these healthcare professionals to meet increased
demand will also be important factors.
One of the alleged shortcomings of user charges for primary
care is that they raise overall costs by causing people to delay seeking care,
leading to worsening of conditions and the subsequent need for more intensive
and expensive care (especially hospitalisation). Certainly, there would seem to
be considerable scope for primary care to prevent hospitalisations. Nineteen per
cent of hospitalisations in New Zealand in 1996 and 1997 were classified as
ambulatory sensitive hospitalisations meaning that they were potentially
avoidable by utilisation of good primary
care.27 The rate of ambulatory sensitive
hospitalisations has been increasing over the last decade.
However, the Rand HIE clearly showed that, for adults at
least, decreased user charges for primary care led to an increased utilisation
of hospitals.10,11 It is certainly not clear
that this should be regarded as a negative effect. It may be that increased
utilisation of primary care allowed the recognition of disease that had been
previously under-treated. It should also be noted again that the Rand HIE
excluded many who might have benefited most from the primary care management of
chronic disease, ie, the elderly and those eligible for Medicaid. Further, the
experiment followed enrolled patients for three or five years only.
A further issue is that the provision of free primary
medical care in New Zealand might encourage a change in style of practice as
well as an improvement in access. Free care would assist the primary care sector
in taking a population and preventive focus. This, in turn, may lead to an
improvement in health and reduced hospitalisation.
In summary, the costs to the Government of providing
‘free’ general practice care are very difficult to estimate with any
accuracy. What is clear is that when trying to estimate these costs it is
important to move beyond simple estimates of how much is currently privately
paid for general practice services. If the primary reason for making general
practice free is to improve access we must expect some increase in utilisation
of these services. Compensation for this increase in utilisation is essential.
In addition, any increase in utilisation will have implications in downstream
costs that are likely to be large. Pharmaceutical, laboratory and other referred
service costs will increase, and it is quite possible that hospital costs will
also increase. Certainly, it would seem to be naive to assume they will
decrease.
Estimates based upon the Rand HIE made in this article vary
widely from $349 million to $922 million depending upon assumptions. A best
estimate would include some degree of limitation on increased utilisation due to
doctors’ responses and a further reduction due to the continued costs of
medicines and other complements. This would imply a total expense to the
government of between $435 and $592 million (scenarios D and B). These costs
need to be balanced against potential benefits, which are not discussed
here.
Author information:
Tom Robinson, Public Health Medicine Registrar, Auckland
Acknowledgements:
Thanks to Toni Ashton and Gregor Coster for their supervision of the
Master’s dissertation upon which this paper is based and their helpful
comments.
Correspondence: Dr
Tom Robinson, 4 Domain Street, Devonport, Auckland. Email: thomas.robinson@xtra.co.nz
References:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|