Elaine Kelly: Growth in NHS-funded elective care
-
Upload
nuffield-trust -
Category
Health & Medicine
-
view
477 -
download
0
Transcript of Elaine Kelly: Growth in NHS-funded elective care
Introduction Background Results Mechanisms Conclusions
More hips, please. Independent sector provision and the
growth in NHS-funded elective care
Elaine Kelly & George StoyeNu�eld Trust Workshop
13th September 2013
1/27
Introduction Background Results Mechanisms Conclusions
Introduction
The past decade of health care policy reforms have increased the role ofcompetition in NHS-funded care.
Existing work has concentrated on the patient choice reforms of 2006and 2008. [Cooper et al, 2011; Gaynor et al, 2012 a,b]
This paper focuses on a separate but related set of reforms thatincreased the access of independent sector providers (ISP) to markets forNHS-funded elective secondary care.
How did this a�ect the market for both NHS and privately funded hipreplacements?
2/27
Figure : Total number of NHS-funded hip replacements in England, by provider type
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
Hip
Re
pla
cem
en
ts
Financial Year
ISP NHS Trusts
The total number of NHS-funded hip replacements increased by 40% between2003/04 and 2010/11.
After 2006/07, most of this growth is accounted for by ISPs.
Figure : Mean hip NHS-funded replacements per Middle Super Output Area bynearest provider type in 2010/11
6
6.5
7
7.5
8
8.5
9
9.5
10
10.5
11
2003/4 2004/5 2005/6 2006/7 2007/8 2008/9 2009/10 2010/11
Me
an h
ip r
ep
lace
men
ts p
er
MO
SA/y
ear
NHS ISP
Growth was fastest in areas where an ISP was located closer than the nearest NHStrust by 2010/11.
Introduction Background Results Mechanisms Conclusions
Research Questions
How did the introduction of ISPs a�ect the market for NHS-funded hipreplacements?
1 Why did the number of hip replacements increase faster in areas whereISPs were located relatively close by?
2 What explains the increase in the number of NHS-funded hipreplacements?
New proceduresSubstitution from privately funded procedures
5/27
Introduction Background Results Mechanisms Conclusions
Independent Sector Provider reforms
1 Independent Sector Treatment Centres (ISTCs)
First introduced in 2003, expanded in 2006.Privately owned but typically treat just NHS-funded patients.Objectives [Naylor & Gregory, 2009]:
Wave 1: to address capacity constraints and reduce waiting timesWave 2: increasing competition for NHS providers, providing more choicesfor patients, and fostering innovation.
2 Any Quali�ed Providers (AQPs)
In mid 2007, choice of providers in orthopaedics expanded to coverexisting facilities, such as private hospitals, through the Extended ChoiceNetwork.Treat privately funded and NHS-funded patients.Extended to other specialties when 2nd choice reform was introduced in2008.
6/27
Figure : NHS-funded hip replacements conducted by ISPs, by quarter and ISP type
050
010
0015
0020
00N
umbe
r IS
P H
ip P
roce
dure
s
2003
q2
2004
q2
2005
q2
2006
q2
2007
q2
2008
q2
2009
q2
2010
q2
2011
q2
Time
ISTC sites AQP sites
ISTC volumes started to increase as ISTCs began to open. Levelled o� after 2008.
AQP volumes increased rapidly after the second choice reform was introduced.
Figure : Number of ISP sites by year and ISP type
0
20
40
60
80
100
120
Sites> 1 pat Sites >20 pats Sites> 1 pat Sites >20 pats
ISTC AQP
Nu
mb
er
of
ISP
Sit
es
2003/4
2004/5
2005/6
2006/7
2007/8
2008/9
2009/10
2010/11
More AQP sites, but ISTC procedures more concentrated across sites.
In 2010/11, average NHS-funded hip replacements per site were 65 for AQPs and
160 for ISTCs.
Figure : Mean number of hip replacements per MSOA/year, by nearest providertype in 2010/11
6
6.5
7
7.5
8
8.5
9
9.5
10
10.5
11
2003/4 2004/5 2005/6 2006/7 2007/8 2008/9 2009/10 2010/11
Me
an h
ip r
ep
lace
me
nts
pe
r M
SOA
/ye
ar
Financial Year
NHS ISTC AQP
Relative growth is much faster in areas where an ISTC is the nearest provider thanwhere AQPs are the nearest provider.
Shift in entire distribution, not just the mean.
Introduction Background Results Mechanisms Conclusions
Why might introducing ISPs a�ect the number of
NHS-funded procedures?
1 Supply: extra potential capacity relaxes supply constraints.
An initial objective of the ISTC programme [Naylor & Gregory, 2009]
2 Demand: ISPs provide an option that potential patients prefer to:
1 No procedure.2 Privately funded treatment
This paper focuses on establishing whether there was a demand response.
Di�culty: all areas/patients can access ISPs through the 2008 choicereforms.
Solution: exploit variation in intensity of treatment or exposure byrelative distance between the nearest ISP and the nearest NHS trust.
10/27
Introduction Background Results Mechanisms Conclusions
Why does the growth rate of hip replacement vary by
distance to ISP?
1 Endogenous placement: ISPs located in areas where higher growth isanticipated/removing supply constraints
2 A demand response:
Patients prefer treatment closer to home [Beckert et al, 2012; Sivey 2012].
Analysis examines ISP placement and the number of hip replacements atthe Middle Super Output Area (MSOA) level.
Data on NHS-funded hip replacements from the inpatient HospitalEpisode Statistics (HES).
6,710 MSOAs in England (ave pop 7,200). MSOAs are a statisticalconstruct, no administrative jurisdictions.
De�ne MSOA as �treated� if there is an ISP that performs hipreplacements nearer than the NHS trust.
11/27
Introduction Background Results Mechanisms Conclusions
What are the determinants of ISP placement?
The odds of MSOA m having an ISP closer than the nearest NHS trust in2010/11 is given by the following speci�cation:
ISPclose10m = θo +θ1WaitTimesm+θ2nTrustm+θ3SDm+ em (1)
WaitTimesm includes waiting times of nearest trust and residents of theMSOA, and MSOA admittances for hip replacements in 2003/04.
nTrustm are characteristics of the nearest trust to MSOA m; SDm aresocio-demographic characteristics (all pre 2005)
Results aim to indicate:
The extent to which ISP placement re�ects population need/supplyconstraints
Any sources of random variation in placement that could be used foridenti�cation
12/27
Table : Odds of having an ISP closer than the nearest NHS trust in 2010/11
Type of ISP Closer than the Nearest Trust
ISP ISP ISP ISTC only AQP only(1) (2) (3) (4) (5)
Nearest Trust Wait Time 2003 (SD) 1.352*** 1.195** 1.141 1.284 1.124(0.102) (0.0900) (0.117) (0.272) (0.128)
MSOA Wait Time 2003 (SD) 0.983 0.963 0.943 0.927 0.955(0.0356) (0.0355) (0.0451) (0.0742) (0.0493)
Average hip replacements in 2003 and 2004 0.972* 0.927*** 0.939** 1.013 0.938**(0.0159) (0.0176) (0.0242) (0.0370) (0.0257)
Distance to Nearest Trust (km) 1.120*** 1.078*** 1.054 1.064**(0.0270) (0.0294) (0.0408) (0.0300)
Distance to Nearest Trust Squared (km) 0.997*** 0.998** 0.999 0.999(0.000760) (0.000768) (0.000941) (0.000789)
IMD score (2004) 0.967*** 0.976 1.063*** 0.941***(0.0117) (0.0179) (0.0185) (0.0198)
Private hospital closer 29.25*** 3.573*** 33.56***(7.843) (1.292) (9.935)
NHS `hospital' (>30 beds) closer 2.028*** 2.146*** 1.915***(0.384) (0.624) (0.386)
Nearest trust & Socio-demographics No Yes Yes Yes Yes
Observations 6,710 6,710 6,710 6,710 6,710Pseudo R-squared 0.0127 0.0731 0.404 0.119 0.413
Notes: *** denotes signi�cance at 1%, ** at 5%, and * at 10% level. Observations are at the MSOA level.
Presence of existing hospital facilities is strongest determinant of ISP location
Adding PCT FE strengthens relationship with private hospital location. (OR increasesto 9.6 in col 4 and 138.7 in col 5)
Introduction Background Results Mechanisms Conclusions
Estimating a demand response
Within PCTs, relative distance to an ISP should not a�ect hipreplacement numbers through supply.
Placement related to nearest trust waiting times but not MSOA waitingtimes (not related to local pre-existing need).Administrative constraints should operate at PCT level, not MSOA level.
However, relative distance to an ISP should a�ect patient demand.
14/27
Table : Treatment and Control Group De�nitions
Financial % MSOA % of MSOA hip rep % ISP patsYear ISP close conducted by ISPs live "close"
ISP closer ISP Further
2003/4 2.7 1.1 0.1 35.02004/5 7.3 2.4 0.9 16.22005/6 8.6 3.9 1.9 17.92006/7 3.6 13.8 3.2 17.42007/8 12.8 13.8 5.6 29.22008/9 19.4 18.1 8.5 36.52009/10 22.3 17.7 10 36.82010/11 28.2 24.4 14.1 45.2
The proportion of areas �treated� by an ISP increases as more ISPs open
Patients are more likely to receive care from an ISP if they live in �treated� areas.
But, most ISP patients do not live in treated areas.
Introduction Background Results Mechanisms Conclusions
Fixed E�ects Speci�cation
Number of residents in MSOA m that receive a NHS-funded hipreplacement (conducted by an NHS trust or an ISP) in year t:
Hipsmt = α +β ISPmt + γm+µt +Xmt + εmt (2)
The coe�cient of interest is β , the e�ect of introducing an ISP close toMSOA m on number of residents admitted for NHS-funded hipreplacements.
Xmt includes time varying MSOA measures of population agecomposition, admissions for fractured neck of femur, and theunemployment rate. εmt clustered at the PCT level.
Identifying assumption: conditional on Xmt , ISPmt uncorrelated with εmt .
16/27
Table : Fixed e�ects estimates of the impact of ISP introduction on number ofadmittances for elective hip replacements per MSOA
Type of ISP Closer: ISP ISTC AQP ISTC20 AQP20(1) (2) (3) (4) (5)
ISP closer than nearest NHS Trust 0.222** 0.447 0.174* 1.189*** 0.825***(0.0983) (0.326) (0.0976) (0.392) (0.168)
Pop 65-79 (thousands) 9.838*** 9.860*** 9.866*** 9.423*** 9.579***(0.867) (0.863) (0.867) (0.860) (0.864)
Pop 80+ (thousands) 9.806*** 9.815*** 9.818*** 9.695*** 9.721***(1.253) (1.255) (1.256) (1.256) (1.264)
FNOF admits 0.0581*** 0.0579*** 0.0581*** 0.0582*** 0.0586***(0.0161) (0.0161) (0.0161) (0.0162) (0.0162)
FNOF admits squared -0.00377*** -0.00375*** -0.00376*** -0.00378*** -0.00377***(0.00123) (0.00124) (0.00123) (0.00123) (0.00123)
Unemployment Rate -8.207 -8.176 -8.214 -9.216 -9.151(6.126) (6.115) (6.134) (6.037) (6.082)
Year Fixed E�ects Yes Yes Yes Yes YesMSOA Fixed E�ects Yes Yes Yes Yes YesDemographics Yes Yes Yes Yes Yes
Observations 46,970 46,970 46,970 46,970 46,970R-squared 0.121 0.121 0.121 0.123 0.124
Notes: *** denotes signi�cance at 1%, ** at 5%, and * at 10% level. Observations are at the MSOA year level. The dependent
variable in all columns is the number of admissions for an NHS-funded elective hip replacement amoungst MSOA residents.
Introduction Background Results Mechanisms Conclusions
Summary
The introduction of ISPs is associated with an increase in demand forhip replacements.
For large ISPs introduced nearer than the nearest trust, ISTCs add 1.2and AQPs 0.8 to annual hip replacements per MSOA.
Relative to a baseline level of hip replacements in 2003/04 of 7.Equivalent to adding an additional 100 people aged 65+ to the MSOApopulation.
Propensity score matching estimates provide a similar set of results.
Potential to use location of existing health care facilities as an IV.
18/27
Introduction Background Results Mechanisms Conclusions
Where is the additional demand for NHS treatment coming
from?
The increase in demand for hip replacements may operate through:
A rise in the number of people having hip replacementsSubstitution from privately funded to NHS-funded hip replacements
Combine HES with hospital level data from the National Joint Registry(NJR), to estimate relationships between NHS, ISP and private payvolumes.
Caution: much more work needed on separating demand from supply.
19/27
Introduction Background Results Mechanisms Conclusions
Data Construction I
HES contains:
Number of patients treated in NHS hospitalsNumber of NHS-funded patients treated in AQPs and ISTCs.
NJR contains:
Number of patient treated in NHS hospitalsTotal number of patients treated in private hospitals, including thoseoperating as AQPs and ISTCs.
Private patients = Hip replacements in private hospitals (NJR) − hipreplacements conducted at ISTCs (NJR)− NHS-funded hip replacementsconducted by AQPs (HES)
Note: measurement error in the number of private procedures.
will be improved with access to patient level data (agreed in principle).
20/27
Introduction Background Results Mechanisms Conclusions
Data Construction II
The NJR has no information on where patients live, therefore assignpatients to areas on the basis of hospital location.
Collapse number of procedures by provider type and NHS/private pay byPrimary Care Trust and �nancial year.
Use data from 2007/08 to 2010/11, due to concerns about quality ofdata in earlier years.
Drop negative private pay volumes.
21/27
Introduction Background Results Mechanisms Conclusions
NJR-HES Sample
Table : PCTs that contain Independent Sector Providers and estimated privatelyfunded hip procedures 2007/08 to 2010/11
PCTs with ISPs No of hip reps on private sitesISTCs AQPs All NHS-funded Est pr pay
NJR HES NJR & HES
2007/8 14 31 18,387 4,222 14,1652008/9 19 48 22,198 6,794 15,4042009/10 20 60 21,511 7,830 13,6812010/11 22 77 22,975 11,665 11,310
Private hospitals treated more patients in 2010/11 than 2007/08.
Increased numbers of NHS-funded patients compensated for falls in private paypatients.
22/27
Introduction Background Results Mechanisms Conclusions
Estimation
Private pay hip replacements and ISTCs
We assume that the supply of ISTC hips is determined by the ISTCcontract and therefore does not respond to private pay volumes.Private pay hip operations in PCT p and �nancial year t is given by:
PPHipspt = α +ρISTCpatspt + γp+µt +Zpt + εpt (3)
Private pay hip replacements and AQPs
We assume that private hospitals strictly prefer to treat private patientsover NHS-funded patients because they receive more for their care.
AQPpatspt = α +σPPHipspt + γp+µt +Zpt + εpt (4)
23/27
Table : Fixed e�ects estimates of the impact of ISP introduction on number ofadmittances for elective hip replacements per PCT of treatment
Priv Funded Ops AQP NHS Ops NHS Trust Ops
(1) (2) (3) (4) (5) (6)
HES ISTC hips -0.155 -0.149 -0.147** -0.144**(0.0995) (0.0958) (0.0628) (0.0618)
Est private pay hips -0.664*** -0.338***(0.141) (0.0842)
HES AQP hips -0.0749 -0.207*(0.0653) (0.109)
Sample All Balanced All Balanced All BalancedPCT & Year FE Yes Yes Yes Yes Yes YesAge Composition Yes Yes Yes Yes Yes Yes
Observations 515 484 515 484 532 520R-squared 0.112 0.165 0.664 0.484 0.043 0.052Number of PCTs 135 121 135 121 136 130
Notes: *** denotes signi�cance at 1%, ** at 5%, and * at 10% level. Observations are at the PCT year level.
Strong evidence of substitution between private pay and AQP procedures, but notbetween private pay and ISTC procedures.
Small negative e�ects of ISTC and AQP procedures on NHS trust procedure numbers
Introduction Background Results Mechanisms Conclusions
Summary
Number of NHS-funded hip replacements increased by 40% between2003/04 and 2010/11, with ISPs accounting for almost two-thirds of therise.
Hip replacements increased faster in areas that were closer to an ISPthan the nearest NHS trust.
Fixed e�ects and matching estimates suggest that this was consistentwith a demand response.
Data on private pay patients from the NJR indicates strong evidence ofsubstitution between private pay and NHS-funded AQP procedures.
Consistent with private hospitals treating NHS patients to helpcompensate for a decline in demand from private patients.
Increases in ISTC and AQP procedures tend to reduce proceduresconducted by NHS trusts.
25/27
Introduction Background Results Mechanisms Conclusions
Implications
1 For patients
ISPs contributed two-thirds of the total increase in hip replacements,contributing a substantive increase in supply.Patients bene�ted more in areas located nearer to an ISP than thenearest trust.
2 For ISPs
ISTC sites provided an unambiguous increase in revenue, as there is notmuch evidence of substitutionFor AQPs, NHS-funded patients have compensated for falls in demandfrom private patients.In the long run could ISPs crowd out private pay patients?
3 For NHS trusts.
There is some evidence that ISP operations led to a fall in NHS trustoperations.Unclear what this means for NHS trust �nances, given likely substitutionto other activity.
26/27
Introduction Background Results Mechanisms Conclusions
Future Work
Add data from 2011/12 and 2012/13.
Patient level data from the National Joint Registry (removing the needto estimate private pay patients).
Use the presence of existing health care facilities as an instrument forISP location.
More theoretical and empirical work separating the supply of health carefrom demand for health care.
27/27