The influence of welfare systems on pay-for-performance programs for general practitioners: A critical review

Accepted Manuscript The influence of welfare systems on pay-for-performance programs for general practitioners: A critical review Mehdi Ammi, Grant Fortier PII:

S0277-9536(17)30108-9

DOI:

10.1016/j.socscimed.2017.02.019

Reference:

SSM 11074

To appear in:

Social Science & Medicine

Received Date: 7 June 2016 Revised Date:

10 February 2017

Accepted Date: 12 February 2017

Please cite this article as: Ammi, M., Fortier, G., The influence of welfare systems on pay-forperformance programs for general practitioners: A critical review, Social Science & Medicine (2017), doi: 10.1016/j.socscimed.2017.02.019. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

The Influence of Welfare Systems on Pay-for-Performance Programs for General Practitioners: A Critical Review

a

RI PT

Mehdi Ammi a,* and Grant Fortier a School of Public Policy and Administration, Carleton University, Ottawa, Ontario, Canada

AC C

EP

TE D

M AN U

SC

* Corresponding author: River Building, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada, K1S 5B6. Email: [email protected]. Phone: +1 (613) 520-2600 x 3227

ACCEPTED MANUSCRIPT

1

The Influence of Welfare Systems on Pay-for-Performance Programs for General

2

Practitioners: A Critical Review

RI PT

3 4

Abstract:

6

While pay-for-performance (P4P) programs are increasingly common tools used to foster quality

7

and efficiency in primary care, the evidence concerning their effectiveness is at best mixed. In

8

this article, we explore the influence of welfare systems on four P4P-related dimensions: the

9

level of healthcare funders’ commitment to P4Ps (by funding and length of program operation),

M AN U

SC

5

program design (specifically target-based vs. participation-based program), physicians’

11

acceptance of the program and program effects. Using Esping-Andersen’s typology, we examine

12

P4P for general practitioners (GPs) in thirteen European and North American countries and find

13

that welfare systems contribute to explain variations in P4P experiences. Overall, liberal systems

14

exhibited the most enthusiastic adoption of P4P, with significant physician acceptance, generous

15

incentives and positive but modest program effects. Social democratic countries showed minimal

16

interest in P4P for GPs, with the exception of Sweden. Although corporatist systems adopted

17

performance pay, these countries experienced mixed results, with strong physician opposition. In

18

response to this opposition, health care funders tended to favour participation-based over target-

19

based P4P. We demonstrate how the interaction of decommodification and social stratification in

20

each welfare regime influences these countries’ experiences with P4P for GPs, directly for

21

funders’ commitment, program design and physicians’ acceptance, and indirectly for program

22

effects, hence providing a framework for analyzing P4P in other contexts or care settings.

AC C

EP

TE D

10

1

ACCEPTED MANUSCRIPT

23 24

Research highlights: •

This study bridges the social policy and physician payment literature

26

•

Welfare systems explain variation in experiences with pay-for-performance

27

•

Liberal systems make enthusiastic use of pay-for-performance (P4P)

28

•

Social democratic systems have minimal interest in P4P for general practitioners

29

•

Opposition in corporatist systems results in participation-based P4P

SC

RI PT

25

M AN U

30 31

Keywords: pay-for-performance; welfare systems; general practitioners; decommodification;

32

social stratification; North America; Europe.

EP AC C

34

TE D

33

2

ACCEPTED MANUSCRIPT

35 36

1. Introduction

Over the last two decades, pay-for-performance (P4P) programs have become popular compensation tools for health care providers. The main objectives of P4P are to improve service

38

quality, contain costs, foster efficiency and ultimately improve patients’ health outcomes.

39

Complementing traditional physician compensation models (e.g. fee-for-service, capitation), P4P

40

programs use financial incentives to explicitly reward adherence to predefined standards of care

41

and cost control (Greene & Nash, 2009). While some early programs penalized health care

42

providers for non-compliance, the vast majority of programs reward physicians in addition to

43

standard compensation (Town, Kane, Johnson, & Butler, 2005). Performance pay initiatives can

44

reinforce multiple dimensions of care, which may be classified under the well-known structure

45

(e.g. electronic health records), process (e.g. disease management for patients with chronic

46

conditions) and outcomes (e.g. increasing the proportion of generic drugs prescribed) framework

47

(Donabedian, 1988). In practice, some P4P programs will reward “participation” rather than

48

“targets” – for instance, if physicians receive bonuses for every additional patient they enroll in a

49

chronic disease management program.

SC

M AN U

TE D

Given the strong impact of general practitioners (GPs) on the utilization of health care

EP

50

RI PT

37

services (namely specialists’ services and prescription drugs), GPs are a natural target for cost-

52

containment strategies including pay-for-performance (Eijkenaar, 2012; Kravet, et al., 2008;

53

Wright & Ricketts, 2010). In addition, quality primary care has been identified as a significant

54

contributor to the overall strength of the healthcare system (Starfield, Shi, & Macinko, 2005),

55

making P4P a strategic component of the quality improvement arsenal.

56 57

AC C

51

Despite the high expectations of its advocates, the literature on the effectiveness of P4P is inconclusive, with findings ranging from a modest improvement in targeted outcomes to

3

ACCEPTED MANUSCRIPT

negligible program impacts (Eijkenaar, 2012; Eijkenaar et al., 2013; Greene & Nash, 2009). This

59

inconclusiveness is attributable to the quality of the research design and statistical analysis used,

60

the variation in individual P4P program design choices (Eijkenaar, 2012; Eijkenaar et al., 2013),

61

as well as difference in contextual and institutional background. The outcomes of P4P also

62

depend on the degree of awareness of the program among physicians (Li et al., 2014), the

63

existence of a dialogue between patients, physicians and payers (Rosenthal & Dudley, 2007) and

64

the pre-existing main payment system (Kantarevic & Kralj, 2013). Nevertheless, a more

65

comprehensive understanding of the factors contributing to P4P effectiveness is necessary,

66

including program acceptance by the providers of care.

SC

M AN U

67

RI PT

58

Eikemo and Bambra (2008) remind us that welfare regimes are an effective way of classifying and understanding the ideological, social and institutional context of health systems

69

in general; thus, they may prove helpful in comprehending the setting within which physician

70

performance pay is implemented. The Esping-Andersen (EA) typology sorts the industrialized

71

nations of Western Europe and North America into three broad welfare state regimes (Esping-

72

Andersen, 1989, 1990): liberal (e.g. USA), corporatist (e.g. Germany) and social-democratic

73

(e.g. Norway). States falling within each regime type are differentiated by (1) the role of the

74

welfare state in altering market forces (decommodification), (2) whether social welfare

75

reinforces class distinctions (social stratification), and (3) the relationship between the state,

76

individual and family (Esping-Andersen, 1989), with membership within each welfare category

77

heavily influenced by the first two attributes of the classification. Although the EA model’s

78

relevance has been criticized (Barrientos, 2008; Orloff, 1993; Van Der Veen & Van Der Brug,

79

2013), it is nevertheless a widely recognized framework that remains central to the health and

80

social policy debates (Emmenegger et al, 2015; Powell & Barrientos, 2015).

AC C

EP

TE D

68

4

ACCEPTED MANUSCRIPT

The goal of this critical literature review is thus to identify whether and how differences

82

between welfare regimes, as identified by Esping-Andersen, affect various P4P program-related

83

dimensions that cover the full program lifecycle: the level of healthcare funders’ (e.g.

84

governments, insurers) commitment to P4Ps (by funding level and length of program operation),

85

program design (specifically target-based vs. participation-based program), the physicians’

86

acceptance of the program (attitudes of general practitioners and their medical association, and

87

participation rates in the program) and program impacts. By examining the variation in P4P

88

adoption between welfare systems rather than examining each program individually, we

89

contribute to the fast growing pay-for-performance literature in a unique way.

SC

M AN U

90

RI PT

81

The paper is organized as follows. We present our methodology and choice of countries in Section 2. Section 3 describes the main P4P initiatives grouped by welfare regime and

92

explores if and how welfare regimes influence experiences with P4P. We discuss these results in

93

Section 4 and conclude in Section 5.

94

2. Methodology

TE D

91

Using a critical literature review approach, we analyzed P4P experiences across welfare

96

systems, with the nation state being the primary unit of analysis. We selected a large sample of

97

countries from each regime for analysis to reduce the risk of selecting outliers in each system

98

category while using the consistency of the results across countries within a typology to ensure

99

the validity of our conclusions. For consistency, we used Esping-Andersen’s (1990) original

AC C

EP

95

100

categories of welfare regimes and their constituent nations (see Table 1). While not all countries

101

fit neatly within the original classification, we retain these groups to ensure a reasonably

102

representative selection of cases across all three categories, and to assess the predictive power of

103

EA’s original typology in terms of P4P experiences.

5

ACCEPTED MANUSCRIPT

104 105

[Table 1 about here]

107

RI PT

106

We conducted a survey of available peer-reviewed and grey English-language P4P

literature using PubMed and Google Scholar, taking an approach similar to a scoping review to

109

map out what is currently known. Firstly, we conducted searches for each country in the study

110

using the following keywords in varying combinations: “pay-for-performance,” “P4P,”

111

“programs,” “healthcare,” “incentives,” “physicians” and “general practitioners.” Secondly, we

112

adopted a snowball strategy by investigating individual sources that were cited in the documents

113

turned up through the initial search, and keeping the referenced document when it provided more

114

detailed information about the experience with P4P in a given country. Finally, the review of

115

international P4P programs by Eijkenaar (2012) identified a number of reliable sources on

116

specific programs and was used for triangulation.

M AN U

TE D

117

SC

108

The goal of these searches was to find evidence of at least one pay-for-performance program active in each country being studied. The program had to be an official initiative in

119

effect at a national or regional/state/county level that provides financial incentives to general

120

practitioners for the meeting of specific “performance” objectives. In the event a nation had

121

multiple P4P programs in place, we selected only one program study based on each program’s

122

significance, how widespread it was amongst physicians and whether sufficient information

123

about the program’s history and effects was available. Note that in some countries, our search

124

failed to turn up any physician pay-for-performance programs. Nevertheless, the absence of P4P

125

in a country is a significant result in itself.

AC C

EP

118

6

ACCEPTED MANUSCRIPT

Having gathered our information, we examined the P4P program literature for four key

127

areas of interest: (1) the level of commitment to the P4P program, as reflected in the program’s

128

start date and the mean proportion of annual physician income derived from P4P participation;

129

(2) physician acceptance of the P4P programs. Indicators include general attitudes of physicians

130

and their representatives (i.e. medical associations) towards proposed P4P programs, and the

131

participation rates of eligible physicians, in cases where participation is voluntary; (3) the

132

program design, particularly as whether the P4P is target-oriented or participation-based; (4) the

133

effects of the programs on physician practice, in terms of measured performance.

SC

M AN U

134

RI PT

126

These outcomes are significant because they are indicative of how the key targets of P4P programs (physicians) will affect (program design) and react to a proposed program before

136

(general attitude and acceptance), during (participation rates and income from performance pay)

137

and after (effects) its implementation, covering the full lifecycle of a given program.

138

3. Results

139

The different outcomes for the P4P programs in each welfare system type are synthetized in

140

Table 2 and discussed hereafter.

142 143 144

EP AC C

141

TE D

135

[Table 2 about here]

7

ACCEPTED MANUSCRIPT

145 146

3.1. Liberal System Physicians and medical associations in liberal regimes are broadly supportive of the concept of P4P and view it as a legitimate policy tool for public and private insurers to improve

148

quality of care and the adoption of preferred practices (Anderson et al., 2006; Cashin, 2014;

149

Cashin & Chi, 2014; Rosenthal & Dudley, 2007). In cases where physicians object to proposed

150

or active P4P initiatives, like in New Zealand, this opposition is typically framed on pragmatic

151

grounds with program design and implementation, rather than moral or ethical opposition to the

152

concept of performance pay (Anderson et al., 2006; Buetow, 2008; Cashin, 2011; Cashin, 2014;

153

Cashin & Chi, 2014; Damberg et al., 2009; McDonald, White & Marmor, 2009). These points of

154

difference (and potentially, resentment) focus on the sizes of P4P incentives as a proportion of a

155

GP’s annual income, the administrative burden imposed by P4P programs, the clinical validity

156

and perceived reasonableness of the measures developed by insurers, as well as whether or not

157

physicians are consulted in this process.

SC

M AN U

TE D

158

RI PT

147

All of the liberal countries researched have pay-for-performance programs active at a national or regional level that were introduced as early as 1998 but no later than 2006. This

160

indicates an earlier adoption of P4P compared to other regimes, while the size of performance

161

incentives in liberal states are relatively high. P4P incentives were as high as 20% in the UK,

162

followed by 10% (Ontario, Canada), and 4-7% (Australia), overall creating significant

163

opportunities for physicians to augment their income by participating in P4P (Buetow, 2008;

164

Cashin, 2014; Cashin, 2011; Cashin & Chi, 2014; Damberg et al., 2009; Greene, 2013; Li et al.,

165

2014; Rosenthal, 2014). Performance income is difficult to determine in New Zealand given the

166

flexible nature of the program (Buetow, 2008; Cashin, 2011). Also, while average rewards are

AC C

EP

159

8

ACCEPTED MANUSCRIPT

167

relatively low in the Californian IHA system, this may conceal significant variation from insurer

168

to insurer in the network of P4P programs (Damberg et al., 2009; Rosenthal, 2014). Interestingly, in all cases the P4P programs are primarily target-based, in that they only

170

reward physicians if they meet specific performance targets, rather than through participation-

171

linked criteria (e.g. the number of patients enrolled in a disease management program). These

172

indicators target areas such as resource use and evidence-based clinical guidelines (Cashin &

173

Chi, 2014; Li et al., 2014; Rosenthal, 2014). This is in contrast to the participation-based

174

programs more common in corporatist countries (see next section).

SC

M AN U

175

RI PT

169

Participation rates in the liberal P4P programs are high, with near universal representation in New Zealand and the UK, and solid majorities in Australia and the US

177

(California) (Buetow, 2008; Cashin, 2011; Cashin, 2014; Cashin & Chi, 2014; Rosenthal, 2014).

178

The lower participation rates in Ontario’s case are partly complicated by the abundance of

179

primary care reforms enacted in the province in the past decade, which have contributed to

180

significant physician confusion and lack of awareness about specific programs available to them

181

(Li et al., 2014).

EP

182

TE D

176

In most programs, physician responded positively to the incentives, resulting in modest but observable improvements along the indicators measured, with the exception of the Australian

184

program. The improvements in physician behaviour were greatest amongst the lowest-

185

performing practices, although a high proportion of GPs already exhibited the recommended

186

behaviours (Latham & Marshall, 2015; Lee et al., 2011). However, modest benefits and a high

187

administrative burden limited the influence of performance pay on physician behaviour (Greene,

188

2013; Kecmanovic & Hall, 2015). This obstacle is exacerbated by the large number of payers

189

and complex payment rules (McDonald, White, & Marmor, 2009; Li, Hurley, Decicca, &

AC C

183

9

ACCEPTED MANUSCRIPT

Buckley, 2014). Moreover, there are indications of performance pay having unintended

191

consequences on physician behaviour in the UK, including evidence of indicator manipulation

192

and weaker performance on non-targeted measures (Doran, et al., 2011; Gravelle, Sutton, & Ma,

193

2010). However, Sutton et al (2010) demonstrate that there could be positive spillovers, where

194

non-targeted indicators improved along with targeted ones in recording patients’ risk factors.

195

RI PT

190

Taken together, these outcomes indicate that funders are more willing to engage in P4P, while physicians working under liberal regimes are more willing to accept, participate in and

197

respond to P4P programs.

198

3.2. Corporatist System

M AN U

199

SC

196

While pay-for-performance has become established in most of the corporatist states studied and will likely remain there for the foreseeable future, the level of support to these

201

programs from physicians and medical associations is mixed at best (Bousquet, Bisiaux, & Chi,

202

2014; Busse, 2004; Chi, 2014; Florentini et al., 2011; Saint-Lary et al., 2013; Schmidt, Tarver, &

203

Ruppe, 2012; van der Heuvel et al., 2010). The strongest adverse reactions of providers to these

204

programs are rooted in fundamental opposition to performance pay in principle. In this context,

205

by forcing P4P on physicians – even if participation is voluntary – healthcare payers are

206

perceived as promoting ethically questionable conduct by creating conflicts of interest between

207

the financial interests of physicians and their patients’ health, as well as infringing on the

208

medical profession’s independence (Bousquet, Bisiaux, & Chi, 2014; Busse, 2004; Chi, 2014;

209

Saint-Lary et al., 2013; Schmidt, Tarver, & Ruppe 2012). These effects were most pronounced in

210

Germany and France. German physicians were concerned about the loss of autonomy in favour

211

of the sickness funds (Busse, 2004; Chi, 2014; van der Heuvel et al., 2010). In France,

212

meanwhile, GPs showed the most overt hostility and voluntary participation was rather low,

AC C

EP

TE D

200

10

ACCEPTED MANUSCRIPT

although the relative popularity of the program eventually encouraged a moderation of the

214

unions’ positions (Bousquet, Bisiaux, & Chi, 2014; Saint-Lary et al., 2013). It is still worth

215

noting that most GPs refused to sign the contracts despite being able to gain financially and

216

withdraw at any time. The intensity of this resistance on the part of medical associations has

217

somewhat softened over time though, as programs became entrenched and take-up gradually

218

increased.

P4P implementation was somewhat delayed in corporatist countries, with some states

SC

219

RI PT

213

starting as late as 2009 and 2011, almost a decade after some liberal regimes, even if some

221

committed earlier. Although there is incomplete information about the sizes of incentives as a

222

proportion of annual physician income, there is evidence that P4P in corporatist countries can

223

amount to a sizable proportion of annual income, at about 5% (Bousquet, Bisiaux, & Chi, 2014;

224

Busse, 2004; Chi, 2014; Florentini et al., 2011; Saint-Lary et al., 2013); this rate is somewhat

225

lower than in liberal countries.

TE D

226

M AN U

220

Contrary to liberal regimes, the majority of the P4P programs in nations with corporatist welfare regimes are participation-based, rather than target-based (Busse, 2004; Chi, 2014;

228

Florentini et al., 2011; Schmidt, Tarver, & Ruppe, 2012). Three of the four states (excepting

229

France) maintain disease management programs that reward physicians in exchange of a

230

responsibility for managing chronically ill patients. This participation-based inclination might

231

have been shaped by the medical community’s initial opposition to P4P.

AC C

232

EP

227

Participation rates of eligible physicians in the corporatist P4Ps are mixed and vary from

233

one jurisdiction to another. The pay-for-participation program in Italy’s Emilia Romagna region

234

had a near universal participation rate, although the rate was much lower for the target-based

235

programs (Florentini et al., 2011). The participation rate in Germany’s DMP is regionally 11

ACCEPTED MANUSCRIPT

variable, while Austria’s only significant P4P program remains fragmented and with low

237

participation (at 15%) (Busse, 2004; Chi, 2014; Schmidt, Tarver, & Ruppe, 2012). Despite this

238

variability, the participation is somewhat limited, notably compared to liberal countries.

239

RI PT

236

Where information about program outcomes is available, corporatist disease management programs have a positive impact on quality of care even though incentives are generally

241

participation-based rather than target-based, although the effectiveness can vary by country

242

(Bousquet, Bisiaux, & Chi, 2014; Busse, 2004; Chi, 2014; Florentini et al., 2011; Or, 2010;

243

Saint-Lary et al., 2013; Saint-Lary & Sicsic, 2015). It is worth noting, however, that changes in

244

physician behaviour resulting from P4P are generally modest.

M AN U

245

SC

240

While there may be clear-cut opposition to payment based on the use of targets, judging by the nature of the corporatist P4P programs currently in use there appears to be greater

247

willingness to accept participation-based pay. Here, GPs may be more likely to perceive

248

themselves as playing a partnership role vis-à-vis the healthcare funders, with the supplementary

249

payment a recognition of the extra work needed to care for chronically ill patients, in contrast to

250

an agreement to be measured and judged for performance against specific indicators.

251

3.3. Social Democratic System

EP

Countries with social democratic regimes have shown a distinct lack of interest in using

AC C

252

TE D

246

253

P4P to reward general practitioners, with the sole exception of Sweden. Finland has no incentive-

254

based programs, and Norway’s commitment to P4P is purely in principle at this point in time

255

(Ringard, et al. 2013; Vuorenkoski, Mladovsky, & Mossialos, 2008). While Sweden and

256

Denmark have implemented P4P programs at the hospital level, only Sweden uses P4P to reward

257

physicians, and the program was established late compared to other systems (Anell and

258

Glenngård 2012; Kristensen et al., 2013; Olejaz et al. 2012). Sweden maintains an amalgam of 12

ACCEPTED MANUSCRIPT

259

P4P initiatives administered at the county-level that provide modest, but still significant

260

incentives (2-4% of annual income) to a number of GPs (Anell & Glenngård, 2012).

261

RI PT

Not surprisingly, there is only limited information on GPs’ participation and the impact of these incentives, and physicians’ attitudes toward this sort of payment is less documented

263

compared to other systems. In public policy circles within these countries, there is increasing

264

acceptance of the idea that provider payment mechanisms should promote quality of care, which

265

may influence attitudes of physicians and make them more conducive to P4P (Ringard, Sagan,

266

Sperre Saunes, & Lindahl, 2013). At minimum, the literature does not indicate any overt

267

opposition to performance pay that has been exhibited in corporatist countries (Anell &

268

Glenngård, 2012; Kristensen, Bech, & Lauridsen, 2013; Olejaz et al., 2012; Ringard et al., 2013).

M AN U

SC

262

269

Obviously, P4P design and effects cannot be judged for social democratic regimes, at the exception of Sweden. However, the extent of physician involvement and participation is

271

unknown there, while the programs are believed to have had only limited impacts on quality of

272

care (Anell and Glenngård 2012; Kristensen et al., 2013; Olejaz et al., 2012; Ringard et al.,

273

2013).

274

3.4. Towards a Theory of How Welfare Systems Influence Experiences with P4P

EP

In summary, clusters of P4P experiences are consistent with the health system categories

AC C

275

TE D

270

276

of the EA framework. Liberal systems have been the most enthusiastic in adopting P4Ps and the

277

programs’ effects have been modestly, though not exclusively, positive. Social democratic

278

countries, on the other hand, have shown a distinct lack of interest in P4P for GPs. Compared to

279

these two welfare system types, the experience of corporatist regimes is somewhat mixed, with

280

participation-based P4P more common there. However, for the EA welfare systems framework

281

to be useful for analysing P4P, we still need to understand how the two important components of 13

ACCEPTED MANUSCRIPT

welfare state typology, social stratification and decommodification, contribute to these results.

283

Figure 1 summarizes findings about P4P experiences in each welfare system (characterized by

284

social stratification and decommodification).

285 286

[Figure 1 about here]

288

SC

287

RI PT

282

Decommodification refers to “the extent to which individuals and families can maintain a normal and socially acceptable standard of living regardless of market performance” (Esping-

290

Andersen, 1990, p. 86). Welfare states modify market forces to different extents:

291

decommodification is high in social-democratic regimes, moderate in corporatists, and low in

292

liberal systems. The level of decommodification in a country may thus have an impact on

293

physicians, whether as employees or contractors of healthcare funders, namely by determining

294

their dependence on the market and interest for market-based mechanisms such as pay-for-

295

performance. Thus, decommodification may substantially influence the level of commitment to

296

P4P within a system, as well as the program design.

EP

TE D

M AN U

289

The welfare state is a determinant of social stratification in that it influences “the

298

articulation of social solidarity, divisions of class, and status differentiation,” (Esping-Andersen,

299

1990, p. 55). Stratification is characterized by occupational distinctions in corporatist regimes,

300

private income as well as stigmatising means-tested assistance in liberal systems, and

301

universalist principles in social democratic regimes. Social stratification may primarily affect

302

physicians’ acceptance of P4P, and in return influence the level of commitment to and program

303

design of the P4P.

AC C

297

14

ACCEPTED MANUSCRIPT

304

The low level of decommodification in liberal regimes leads to physicians being highly dependent on the market (given the relative lack of social supports), and in turn favourable to

306

pay-for-performance programs that provide a clear boost to their incomes. This has been the

307

experience in almost all the liberal countries surveyed, with the exception of New Zealand,

308

although incentives were significantly lower in that case (Buetow, 2008; Cashin, 2011). In this

309

vein, it is no surprise that performance pay was introduced early on and that program spending

310

was high in liberal systems (up to 20% of physician income in the UK, Cashin, 2014). Moreover,

311

P4P is consistent with the higher tolerance for market-based stratification. P4P incentives are

312

thus likely to be perceived as legitimate rewards provided in exchange for higher quality or more

313

efficient care, paving the way for the target-based programs that dominate liberal regimes.

SC

M AN U

314

RI PT

305

Even after accounting for physicians’ status as self-governing medical professionals across welfare regimes (Timmermans & Oh, 2010), due to strong occupational stratification,

316

physicians in corporatist nations are more conscious of themselves as a unified social class than

317

in other welfare systems. This distinction is clear in France and Germany, were medical

318

associations and a significant portion of their members have strongly opposed the introduction of

319

P4P by the health funders (Bousquet, Bisiaux, & Chi, 2014; Busse, 2004; Chi, 2014; Saint-Lary

320

et al., 2013). The moderate commodification within corporatist regimes further reinforces

321

resistance to attempts to “commoditize” the medical profession with market mechanisms such as

322

P4P, even if individual professionals may stand to gain financially. As a result, physicians may

323

have been able to leverage on their influence to ensure that the design of the P4P is participation-

324

based rather than target-based, in all but one country.

325 326

AC C

EP

TE D

315

The near-absence of P4P for general practitioners in social democratic system is consistent with the underlying universalist stratification where there is little appetite for and

15

ACCEPTED MANUSCRIPT

commitment to interventions that are likely to generate large differences in physician income

328

equivalent to the levels seen in liberal countries (Fujusawa & Lafortune, 2008). Even the

329

Swedish program, the only one in effect, provides only minimal rewards (Anell & Glenngård,

330

2012). The effect of the universalist stratification is reinforced by the high decommodification,

331

such that the social democratic systems exhibit a lack of interest for, or at least of use of, market-

332

driven instruments like P4P.

Social stratification, decommodification, and their interaction within a welfare system

SC

333

RI PT

327

affect the level of commitment, program design and physician acceptance of P4P programs. The

335

framework in Figure 2 illustrates the feedback loops between physician acceptance and program

336

design, as well as with the level of commitment to the program. These feedback loops are well

337

illustrated in France, where the national health insurer initiated a timid program (program design

338

and level of commitment) that encountered fierce opposition and low take-up (physician

339

acceptance). The insurer later revised the program (program design) in negotiations with the

340

medical associations to develop more ambitious and comprehensive targets (level of

341

commitment) (Bousquet, Bisiaux, & Chi, 2014; Saint-Lary et al., 2013).

TE D

EP

342

M AN U

334

Figure 2 also shows that while the interaction of social stratification and decommodification directly affects commitment, design and acceptance of P4Ps, the welfare

344

system has an indirect impact on the program effects on GPs’ behaviour. This last effect is thus

345

likely to be marginal. Indeed, comparing the effects of P4P programs between liberal and

346

corporatist regimes show that, even if P4P are more successful in changing physicians’

347

behaviour toward targeted outcomes in liberal systems than in corporatist ones, the actual

348

changes in measured performance remain modest overall (see Table 2).

AC C

343

349

16

ACCEPTED MANUSCRIPT

350 351

[Figure 2 about here]

4. Discussion

354

4.1. Policy Implications

355

SC

353

RI PT

352

While P4P programs were greeted with enthusiasm by many healthcare funders, they might not fit all health systems. Nevertheless, pay-for-performance has become entrenched in

357

most of the countries studied (with the notable exception of social democratic regimes) and the

358

overall trend is towards expanding, rather than reducing the role of P4P as a physician payment

359

mechanism (Eijkenaar, 2012). However, the implications for policy makers in liberal and

360

corporatist regimes are different.

TE D

M AN U

356

Healthcare funders considering new P4P programs need to anticipate the types of

362

negative responses that proposals are likely to garner from physicians and medical associations.

363

In liberal nations, low decommodification and income stratification create ideal conditions for

364

P4P and healthcare funders will not face principles-based opposition from medical associations.

365

Even in New Zealand, medical association opposition concentrated on the perceived low

366

rewards, rather than ethical concerns. In these regimes, policy-makers will need to prioritize

367

program design and implementation issues (e.g. validity and credibility of performance

368

measures, amount of payment). The visibility of the program and the timing of its introduction

369

are crucial points for liberal systems policy-makers. The Canadian experience makes clear that

370

multiple concurrent reforms will impede P4P program success, but ensuring the P4P salience

371

might prove difficult in the US context of multiple-payers. Experience with the Quality and

AC C

EP

361

17

ACCEPTED MANUSCRIPT

Outcome Framework in the UK warns policymakers against the difficulties of changing

373

performance targets midstream, and the need for encouraging careful reflections on the program

374

design early on.

RI PT

372

In cases where opposition to P4P is on ethical, principles-based grounds like in

376

corporatist regimes, it is especially important for healthcare funders to justify the legitimacy of

377

their P4P proposals first, before more “practical” issues of program design can be negotiated and

378

resolved with the medical associations. In regimes characterized by a strong social stratification

379

and moderate decommodification, policy-makers need to consider physicians as equal partners in

380

the P4P elaboration. Indeed, as shown by the Austrian experience, the tension between health

381

funder and medical representative can be eased, however, the adopted P4P is then more likely to

382

be participation-based. In Italy, where both participation-based and target-based P4P have been

383

in place, physicians’ participation in the target-based program has been dismal (see Table 2).

384

German experimental evidence indicates that physicians would expect to be heavily

385

compensated to participate in target-based P4P (Krauth, Liersch, Jensen, & Amelung, 2016). It

386

might thus be the case that target-based P4P programs are not suitable for corporatist systems.

387

The only country considered that has an active target-based program (France) did not experience

388

meaningful performance improvement. Therefore, other ways to incentivize physicians’

389

behaviour toward more quality or cost-containment could be mobilized in corporatist regimes.

390

Non-financial interventions, such as continuing education, clinical guidelines and performance

391

feedback (Dexheimer et al., 2008; Farmer et al., 2008; Forsetlund et al., 2009) might generate

392

less opposition from the medical profession in this regime and be more effective at changing

393

GPs’ behaviour. Indeed, and as an illustration, French general practitioners clearly prefer these

394

non-financial interventions to the French P4P program (Ammi & Peyron, 2016).

AC C

EP

TE D

M AN U

SC

375

18

ACCEPTED MANUSCRIPT

395

Drawing clear implications for the social democratic regimes is not possible as Sweden is the only country with an active P4P for GPs. Since the lack of interest for P4P in social

397

democratic regimes is established, policy-makers in these countries could usefully mobilize non-

398

financial interventions, in lieu of P4P, to attain their quality or efficiency objectives.

399

RI PT

396

When exploring the effects of P4P across welfare systems, it is important to note the limited exploration of unintended consequences of these programs. A large body of literature in

401

social psychology and economics warns about the potential crowding-out effects of extrinsic

402

motivations (e.g. bonuses) on intrinsic ones (Benabou & Tirole, 2003; Deci, 1971; Frey & Jegen,

403

2001). In the instance of P4P for GPs, the (chiefly British) literature shows that the negative

404

effects should not be underestimated (Doran, et al., 2011; Gravelle, Sutton, & Ma, 2010;

405

McDonald & Roland, 2009). We did not identify any studies that specifically explored negative

406

spillovers outside of liberal nations, hence we cannot directly comment on P4P’s unintended

407

consequences in comparison to other regimes. However, beyond the requirement for more

408

empirical research, this finding underlines the need for policy-makers to design information

409

systems that monitor both targeted and non-targeted indicators to ensure the absence of negative

410

spillovers of their P4P program, or correct them in case they happen.

EP

TE D

M AN U

SC

400

412 413

AC C

411

4.2. Limitations

The Esping-Andersen typology is not without its criticisms and the empirical validity of

414

its country clustering has been questioned (Scholnick, 2005; Van der Veen and Van der Brug,

415

2012). While further works have expanded on (Barrientos, 2008) or amended (Orloff, 1993)

416

Esping-Andersen’s typology, none has yet succeeded in replacing it as the leading frame of

417

reference for studying social policy in developed countries. Indeed, the concepts of 19

ACCEPTED MANUSCRIPT

decommodification and stratification used to distinguish regimes remain central contributions to

419

the welfare state literature (Arts and Gelissen, 2002; Emmenegger et al., 2015; Powell and

420

Barrientos, 2015). The EA framework continues to predict phenomena such as fertility rates

421

(McDonald & Moyle, 2010), and our findings indicate that EA framework effectively explains

422

variations in national P4P experiences. While other healthcare-related typologies exist, including

423

several with an explicit focus on healthcare service delivery and financing (Bohm, et al. 2013;

424

Wendt, Frisina and Rothgang 2009), the major issue with these typologies is endogeneity, since

425

physician payment is one of the variables used to construct the typology. Moreover, these

426

typologies cannot explain how the physicians and the medical community – as policy actors in a

427

complex social, cultural and institutional environment – may react to P4P. Overall, we believe

428

that the choice of Esping-Andersen model was the most appropriate one, and we have shown not

429

only that this framework is robust to P4P experiences, but also how the interaction of social

430

stratification and decommodification provide clear insights on how to interpret empirical

431

clusters.

SC

M AN U

TE D

432

RI PT

418

Our critical review of international P4P programs was however limited to Englishlanguage literature, which may cause some publication bias towards information sources from

434

nations with predominantly Anglophone populations, which are all designated as liberal nations

435

by Esping-Andersen. It is possible that additional sources about programs in the non-English-

436

speaking world, which could have supported or moderated the study results, may have been

437

available only in their local languages. Moreover, there are gaps in the information available

438

about individual programs, such as participation rates and the proportion of annual income

439

derived from P4P. Some programs have been more thoroughly and recently studied than others,

440

thus leading to variations in the data available about certain programs. However, to the best of

AC C

EP

433

20

ACCEPTED MANUSCRIPT

our knowledge, this is the first study that bridges the social policy and physician payment

442

literature, hence articulating the empirical findings of the pay-for-performance literature within a

443

respected theoretical framework.

444

RI PT

441

It might be argued that social democratic countries have made lower commitments to P4P because there is little perceived need for it, notably due to already high quality of care and

446

efficiency. In fact, this is not the case as these nations are ranked very low on their ability to

447

provide effective, safe, coordinated and patient-centred care, compared to their developed

448

country peers (Davis, Stremikis, Squires, & Schoen, 2014). Furthermore, all OECD member

449

states (including social democratic countries) have the potential to achieve significant savings

450

through more cost efficient care without compromising population health outcomes (OECD,

451

2010). We thus believe that the universalist environment contributes to the lack of interest in P4P

452

more than the health care system performance itself.

M AN U

SC

445

Another limitation of this study is the lack of direct generalizability outside the

454

considered countries and primary care setting. We elected to analyze only the countries present

455

in Esping-Andersen’s original typology. However, P4P programs for general practitioners are in

456

effect elsewhere in the developed world (Cashin et al., 2014). Even low- and middle-income

457

countries are increasingly adopting P4P in this context (Miller & Babiarz, 2014). Even though

458

we focused on primary care, as GPs are a target of choice for P4P, this payment scheme is also

459

used in inpatient care (Cashin et al., 2014). It is not immediately clear if our findings would

460

apply directly to the hospital setting. The incentive structure within hospital differs markedly

461

from ambulatory care and there is a large variation of P4P programs in terms of design and target

462

in hospitals (Milstein & Schreyogg, 2016). While findings from our study cannot directly inform

463

P4P experiences in other settings, we highlighted the key role of social stratification and

AC C

EP

TE D

453

21

ACCEPTED MANUSCRIPT

decommodification in explaining P4P experiences in a framework summarized in Figure 2.

465

Further research could use this framework and explore its robustness in other countries where a

466

welfare state is present or nascent, particularly East Asian, Latin American and Eastern European

467

countries, or in the hospital setting, where many OECD countries have an active P4P program in

468

the inpatient sector.

469

RI PT

464

Although welfare systems directly affect the level of commitment of health funders, physicians’ acceptance of P4P programs and whether the program is target-based vs.

471

participation-based, they only have an indirect impact on P4P program effects. When it comes to

472

physician response to a P4P, program design choices matter – for instance, the encouraged

473

behaviours, the performance indicators used to measure them, the use of absolute vs. relative

474

targets, the frequency of payment, and the use of a risk adjustment formula (Eijkenaar, 2013).

475

While we have been able to identify and explain the dominance of target-based P4P in liberal

476

systems and participation-based ones in corporatist regimes, we did not find any systematic

477

association between other elements of program design and welfare systems. Similarly, Eijkenaar

478

(2012) noted the arbitrary character of P4P program design choices. The clinical validity of the

479

indicators used, the attainability of the performance targets set are some of the key factors that

480

can contribute to the success of a P4P program (Eijkenaar, 2013). To fully comprehend P4P in

481

primary care, understanding what factors trigger differences in program design choices (beyond

482

the target vs. participation base) is an important endeavor.

M AN U

TE D

EP

AC C

483

SC

470

22

ACCEPTED MANUSCRIPT

484 485

5. Conclusion Overall, our findings demonstrate that Esping-Andersen’s typology of welfare systems effectively explains national P4P experiences. Liberal regimes exhibit the most enthusiastic

487

usage of pay-for-performance programs, with significant medical association support, high

488

participation rates, early implementation, generous incentives, and target-based rewards.

489

Corporatist regimes all adopted P4P and rely heavily on participation-based models, but their

490

experiences are mixed due to variable participation as well as fierce physician opposition.

491

Finally, social democratic regimes have no P4P programs providing rewards to general

492

practitioners, with the exception of Sweden, indicating that social democratic nations have low

493

interest in P4P initiatives.

SC

M AN U

494

RI PT

486

Furthermore, we establish a framework where the effect of the interaction of social stratification and decommodification within a nation explains how the P4P experiences conform

496

to the categories developed by Esping-Andersen. This framework highlights the direct influence

497

of welfare systems on healthcare funders’ commitment, target-oriented vs. participation-based

498

program design, physician acceptance of P4Ps, as well as the indirect influence on P4P effects on

499

physician performance. It could be used to analyze P4P in other contexts or care settings.

500

In order to synthesize international evidence, many systematic reviews of the P4P

AC C

EP

TE D

495

501

literature, and even systematic review of systematic reviews (Eijkenaar et al., 2013) have been

502

realised. We have shown that the welfare system type, via social stratification and

503

decommodification, has an influence on multiple dimensions of P4P. Although data availability

504

could pose a challenge, a quantitative meta-analysis of the available evidence could more

505

precisely quantify these influences.

506

23

ACCEPTED MANUSCRIPT

507 508 509 510

Bibliography

511 512 513

Anderson, K., Sebaldt, R., Lohfeld, L., Burgess, K., Donald, F., & Kaczorowski, J. (2006). Views of family physicians in southwestern Ontario on preventive care services and performance incentives. Family Practice, 23, 469-471. doi:PMC3114679

514 515 516

Anell, A., & Glenngård, A. M. (2012). Sweden: Health system review. Health Systems in Transition, 14(5), 1-159. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0008/164096/e96455.pdf

517 518

Arts, W., & Gelissen, J. (2002). Three worlds of welfare capitalism or more? A state-of-the-art report. Journal of European Social Policy, 12(2), 137-158.

519 520

Bambra, C. (2005, April). Cash Versus Services: 'Worlds of Welfare' and the Decommodification of Cash Benefits and Health Care. Journal of Social Policy, 2, 195-213. doi:10.1017/S0047279404008542

521 522

Barrientos, A. (2008). Labour markets and the (hyphenated) welfare regimes in Latin America. Economy and Society, 38(17), 87-108.

523 524

Benabou, R., & Tirole, J. (2003). Intrinsic and Extrinsic Motivation. Review of Economic Studies, 70, 489520.

525 526 527

Bohm, K., Schmid, A., Gotze, R., Landwehr, C., & Rothgang, H. (2013). Five types of OECD healthcare systems: Empirical results of a deductive classification. Health Policy, 113, 258-269. doi:10.1016/j.healthpol.2013.09.003

528 529 530 531 532 533

Bousquet, F., Bisiaux, R., & Chi, Y.-L. (2014). France: Payment for public health objectives. In E. O. Policies, C. Cashin, Y.-L. Chi, P. Smith, M. Borowitz, & S. Thomson (Eds.), Paying for Performance in Health Care: Implications for health system performance and accountability. New York: Open University Press. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-for-Performance-inHealth-Care.pdf

534 535

Buetow, S. (2008). Pay-for-performance in New Zealand primary health care. Journal of Health Organization and Management, 22(1), 36-47. doi:PMID: 18488518

536 537

Busse, R. (2004). Disease Management Programs in Germany's Statutory Health Insurance System. Health Affairs, 23(3), 56-67. doi:PMID: 15160803

538 539

Cashin, C. (2011). New Zealand: Primary Health Organization (PHO). World Bank. Retrieved from http://www.rbfhealth.org/sites/rbf/files/Case%20study%20New%20Zealand%20PHO.pdf

540 541 542 543

Cashin, C. (2014). United Kingdom: Quality and outcomes framework. In E. O. Policies, C. Cashin, Y.-L. Chi, P. Smith, M. Borowitz, & S. Thomson (Eds.), Paying for Performance in Health Care: Implications for health system performance and accountability (pp. 205-221). New York: Open University Press. Retrieved from

AC C

EP

TE D

M AN U

SC

RI PT

Ammi, M., & Peyron, C. (2016). Heterogeneity in general practitioners’ preferences for quality improvement programs: a choice experiment and policy simulation in France. Health Economics Review, 6(44). doi:10.1186/s13561-016-0121-7

24

ACCEPTED MANUSCRIPT

544 545

http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-for-Performance-inHealth-Care.pdf Cashin, C., & Chi, Y.-L. (2014). Australia: Practice incentives programme. In E. O. Policies, C. Cashin, Y.-L. Chi, P. Smith, M. Borowitz, & S. Thomson (Eds.), Paying for Performance in Health Care: Implications for health system performance and accountability (pp. 109-125). New York: Open University Press. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-for-Performance-inHealth-Care.pdf

552 553 554 555 556

Cashin, C., Chi, Y.-L., Smith, P., Borowitz, M., & Thomson, S. (Eds.). (2014). Paying for Performance in Health Care: Implications for health system performance and accountability. New York: Open University Press. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-for-Performance-inHealth-Care.pdf

557 558 559 560 561

Chi, Y.-L. (2014). Germany: Disease management programmes. In E. O. Policies, C. Cashin, Y.-L. Chi, P. Smith, M. Borowitz, & S. Thomson (Eds.), Paying for Performance in Health Care: Implications for health system performance and accountability (pp. 157-171). New York: Open University Press. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-forPerformance-in-Health-Care.pdf

562 563

Damberg, C., Raube, K., Teleki, S., & dela Cruz, E. (2009). Taking Stock of Pay-for-Performance: A candid assessment from the front lines. Health Affairs, 28(2), 517-525. doi:10.1377/hlthaff.28.2.517

564 565 566 567

Davis, K., Stremikis, K., Squires, D., & Schoen, C. (2014). Mirror, Mirror on the Wall: How the Performance of the U.S. Health Care System Compares Internationally. The Commonwealth Fund. Retrieved from http://www.commonwealthfund.org/~/media/files/publications/fundreport/2014/jun/1755_davis_mirror_mirror_2014.pdf

568 569

Deci, E. (1971). Effects of externally mediated rewards on intrinsic motivation. Journal of Personality and Social Psychology, 18, 105-115.

570 571 572

Dexheimer, J., Talbot, T., Sanders, D., Rosenbloom, S., & Aronsky, D. (2008). Prompting Clinicians about Preventive Care Measures: A Systematic Review of Randomized Controlled Trials. Journal of the American Medical Informatics Association, 15(3), 311-320.

573 574

Donabedian, A. (1988). The quality of care. How can it be assessed? Journal of the American Medical Association, 260(12), 1743-1748. doi:10.1001/jama.260.12.1743

575 576 577 578

Doran, T., Kontopantelis, E., Valderas, J. M., Campbell, S., Roland, M., Salisbury, C., & Reeves, D. (2011). Effect of financial incentives on incentivised and non-incentivised clinical activities: longitudinal analysis of data from the UK Quality and Outcomes Framework. BMJ, 342, d3590. doi:10.1136/bmj.d3590

579 580

Eijkenaar, F. (2012). Pay for Performance in Health Care: An International Overview of Initiatives. Medical Care Research and Review, 69(3), 251–276. doi:10.1177/1077558711432891

AC C

EP

TE D

M AN U

SC

RI PT

546 547 548 549 550 551

25

ACCEPTED MANUSCRIPT

Eijkenaar, F. (2013). Key issues in the design of pay for performance programs. European Journal of Health Economics, 14(1), 117-131. doi:10.1007/s10198-011-0347-6

583 584 585

Eijkenaar, F., Emmert, M., Scheppach, M., & Schoffski, O. (2013). Effects of pay for performance in health care: A systematic review of systematic reviews. Health Policy, 110, 115-130. doi:10.1016/j.healthpol.2013.01.008

586 587

Eikemo, T., & Bambra, C. (2008). The welfare state: a glossary for public health. Journal of Epidemiology and Community Health, 62(1), 3-6. doi:10.1136/jech.2007.066787

588 589 590

Emmenegger, P., Kvist, J., Marx, P., & Petersen, K. (2015, February 26). Special Issue: 25 Years of 'Three Worlds of Welfare Capitalism'. Journal of European Social Policy, 25(1). Retrieved from http://www.socialsciencespace.com/2015/02/25-years-of-three-worlds-of-welfare-capitalism/

591 592

Esping-Andersen, G. (1989). The three political economies of the welfare state. Canadian Review of Sociology and Anthropology, 26(1), 10-36.

593

Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism. Cornwall, UK: TJ Press.

594 595 596

Farmer, A., Légaré, F., Turcot, L., Grimshaw, J., Harvey, E., McGowan, J., & Wolf, F. (2008, July 16). Printed educational materials: effects on professional practice and health care outcomes. Cochrane Database of Systematic Reviews(3). doi:CD004398

597 598 599

Florentini, G., Iezzi, E., Lippi Bruni, M., & Ugolini, C. (2011). Incentives in primary care and their impact on potentially avoidable hospital admissions. The European Journal of Health Economics, 12(4), 297-309. doi:10.1007/s10198-010-0230-x

600 601 602

Forsetlund, L., Bjørndal, A., Rashidian, A., Jamtvedt, G., O'Brien, M., Wolf, F., . . . Oxman, A. (2009, April 15). Continuing education meetings and workshops: effects on professional practice and health care outcomes. Cochrane Database of Systematic Reviews(2). doi:CD003030

603 604

Frey, B., & Jegen, B. (2001). Motivation Crowding Theory: A Survey of Empirical Evidence. Journal of Economic Surveys, 15, 589-611.

605 606 607

Fujisawa, R., & Lafortune, G. (2008). The Remuneration of General Practitioners and Specialists in 14 OECD Countries: What are the Factors Influencing Variations across Countries? OECD Health Working Papers(41). doi:10.1787/228632341330

608 609 610

Gravelle, H., Sutton, M., & Ma, A. (2010). Doctor Behaviour under a Pay for Performance Contract: Treating, Cheating and Case Finding? The EconomicJournal, 120(542), F129-F156. doi:10.1111/j.1468-0297.2009.02340.x

611 612

Greene, J. (2013, August). An Examination of Pay-for-Performance in General Practice in Australia. Health Services Research, 48(4), 1415-1432. doi:10.1111/1475-6773

613 614

Greene, S., & Nash, D. (2009, March/April). Pay for Performance: An Overview of the Literature. American Journal of Medical Quality, 24(2), 140-163. doi:10.1177/1062860608326517

615 616 617

Kantarevic, J., & Kralj, B. (2013). Link between Pay for Performance Incentives and Physician Payment Mechanisms: Evidence from the Diabetes Management Incentive in Ontario. Health Economics, 22, 1417-1439. doi:10.1002/hec.2890

AC C

EP

TE D

M AN U

SC

RI PT

581 582

26

ACCEPTED MANUSCRIPT

Kecmanovic, M., & Hall, J. (2015, May). The use of financial incentives in Australian general practice. The Medical Journal of Australia, 202(9), 488-492. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/25971573

621 622 623

Krauth, C., Liersch, S., Jensen, S., & Amelung, V. E. (2016). Would German physicians opt for pay-forperformance programs? A willingness-to-accept experiment in a large general practitioners’ sample. Health Policy, 120(2), 148-158. doi:10.1016/j.healthpol.2016.01.009

624 625 626

Kravet, S., Shore, A., Miller, R., Green, G., Kolodner, K., & Wright, S. (2008, February). Health Care Utilization and the Proportion of Primary Care Physicians. American Journal of Medicine, 121(2), 142-148. doi:10.1016/j.amjmed.2007.10.021

627 628 629 630

Kristensen, S., Bech, M., & Lauridsen, J. (2013). Who to pay for performance? The choice of organization level for hospital performance incentives. Centre of Health Economics Research, University of Southern Denmark. Retrieved from http://www.sdu.dk//media/files/om_sdu/centre/cohere/working+papers/2013/2013_5.pdf

631 632

Latham, L., & Marshall, E. (2015). Performance-Based Financial Incentives for Diabetes Care: An Effective Strategy? Canadian Journal of Diabetes, 39, 83-87. doi:10.1016/j.jcjd.2014.06.002

633 634 635

Lee, J., Netuveli, G., Majeed, A., & Millett, C. (2011, December). The Effects of Pay for Performance on Disparities in Stroke, Hypertension, and Coronary Heart Disease Management: Interrupted Time Series Study. PLoS one, 6(12), 1-8. doi:10.1371/journal.pone.0027236

636 637

Li, J., Hurley, J., Decicca, P., & Buckley, G. (2014). Physician Response to Pay-for-Performance: Evidence from a natural experiment. Health Economics, 23, 962-978. doi:10.1002/hec.2971

638 639 640

Lo Scalzo, A., Donatini, A., Orzella, L., Cicchetti, A., Profi li, S., & Maresso, A. (2009). Italy: Health system review. Health Systems in Transition, 11(6), 1-216. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0006/87225/E93666.pdf

641 642

McDonald, P., & Moyle, H. (2010). Why do English-speaking countries have relatively high fertility? Journal of Population Research, 27, 247-273. doi:10.1007/s12546-010-9043-0

643 644 645

McDonald, R., & Roland, M. (2009). Pay for performance in primary care in England and California: a comparison of unintended consequences. Annals of Family Medicine, 7(2), 121-127. doi:10.1370/afm.946

646 647 648

McDonald, R., White, J., & Marmor, T. (2009, October). Paying for Performance in Primary Medical Care: Learning about and Learning from "Success" and "Failure" in England and California. Journal of Health Politics, Policy and Law, 34(5), 747-776. doi:doi: 10.1215/03616878-2009-024

649 650 651

Miller, G., & Babiarz, K. (2014). Pay-for-Performance Incentives in Low- and Middle-Income Country Health Programs. Encyclopedia of Health Economics, 457-466. doi:10.1016/B978-0-12-3756787.00126-7

652 653 654

Milstein, R., & Schreyogg, J. (2016). A review of pay-for-performance programs in the inpatient sector in OECD countries. Health Policy, 120(10), 1125-1140. doi:http://dx.doi.org/10.1016/j.healthpol.2016.08.009

AC C

EP

TE D

M AN U

SC

RI PT

618 619 620

27

ACCEPTED MANUSCRIPT

OECD. (2010). Health care systems: Getting more value for money. OECD Economics Department Policy Notes(2), 1-11.

657 658 659

Olejaz, M., Juul Nielsen, A., Rudkjøbing, A., Okkels Birk, H., Krasnik, A., & Hernández-Quevedo, C. (2012). Denmark: Health system review. Health Systems in Transition, 14(2), 1-192. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0004/160519/e96442.pdf

660 661

Or, Z. (2010). P4P for generalists: first results. Retrieved January 2015, from Health Policy Monitor: http://hpm.org/en/Surveys/IRDES_-_France/16/P4P_for_generalists__first_results.html

662 663

Orloff, A. S. (1993). Gender and the social rights of citizenship: State policies and gender relations in comparative perspective. 58(3), 303-328.

664 665

Powell, M., & Barrientos, A. (2015). Introduction: Twenty-Five Years of the Welfare Modelling Business. Social Policy and Society, 14(2), 241-245. doi:10.1017/S147474641400058X

666 667 668

Ringard, Å., Sagan, A., Sperre Saunes, I., & Lindahl, A. (2013). Norway: Health system review. Health Systems in Transition, 15(8), 1-162. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0018/237204/HiT-Norway.pdf?ua=1

669 670

Rosenthal, B., & Dudley, R. (2007). Pay-for-Performance: Will the Latest Payment Trend Improve Care? Journal of the American Medical Association, 297(7), 740-743. doi:PMID: 17312294

671 672 673 674 675 676

Rosenthal, M. (2014). United States: California integrated healthcare association physician incentive program. In C. Cashin, Y.-L. Chi, P. Smith, M. Borowitz, & S. Thomson (Eds.), Paying for Performance in Health Care: Implications for health system performance and accountability (pp. 223-236). New York: Open University Press. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0020/271073/Paying-for-Performance-inHealth-Care.pdf

677 678

Saint-Lary, O., & Sicsic, J. (2015). Impact of a pay for performance programme on French GPs' consultation length. Health Policy, 119, 417-426. doi:10.1016/j.healthpol.2014.10.001

679 680 681

Saint-Lary, O., Bernard, E., Sicsic, J., Plu, I., Francois-Purssell, I., & Franc, C. (2013, September). Why did most French GPs choose not to join the voluntary national pay-for-performance program? PLoS ONE, 8(9). doi:10.1371/journal.pone.0072684

682 683 684

Schmidt, A., Tarver, L., & Ruppe, G. (2012). Innovations in Health Care Quality Measurement and Management in Austria. Vienna: European Centre for Social Welfare Policy and Research. Retrieved from http://www.euro.centre.org/data/1384355081_34166.pdf

685 686

Scholnick, E. (2005). Is Esping-Andersen's Liberal Regime Adequate: Health Care in the United States, United Kingdom and Canada. Western Political Science Association Conference.

687 688 689

Starfield, B., Shi, L., & Macinko, J. (2005, September). Contribution of Primary Care to Health Systems and Health. The Milbank Quarterly, 83(3), 457-502. Retrieved from http://www.commonwealthfund.org/usr_doc/Starfield_Milbank.pdf

AC C

EP

TE D

M AN U

SC

RI PT

655 656

28

ACCEPTED MANUSCRIPT

Sutton, M., Elder, R., Guthrie, B., & Watt, G. (2010). Record rewards: the effects of targeted quality incentives on the recording of risk factors by primary care providers. Health Economics, 19(1), 113. doi:10.1002/hec.1440

693 694

Timmermans, S., & Oh, H. (2010). The Continued Social Transformation of the Medical Profession. American Sociological Association, 51(S), S94-S106. doi:10.1177/0022146510383500

695 696 697

Town, R. J., Kane, R., Johnson, P., & Butler, M. (2005). Economic Incentives and Physicians’ Delivery of Preventive Care: A Systematic Review. American Journal of Preventive Medicine, 28(2), 234-240. doi:10.1016/j.amepre.2004.10.013

698 699 700

van der Heuvel, H., Mand, P., Heim, S., & Hummers-Pradier, E. (2010). Views of German general practitioners on the clinical indicators of the British Quality and Outcomes Framework. Quality in Primary Care, 18, 85-92. doi:PMID: 20529469

701 702 703

Van Der Veen, R., & Van Der Brug, W. (2013, April). Three Worlds of Social Insurance: On the Validity of Esping-Andersen's Welfare Regime Dimensions. British Journal of Political Science, 43(02), 323343. doi:10.1017/S0007123412000257

704 705 706

Vuorenkoski, L., Mladovsky, P., & Mossialos, E. (2008). Finland: Health system review. Health Systems in Transition, 10(4), 1-168. Retrieved from http://www.euro.who.int/__data/assets/pdf_file/0007/80692/E91937.pdf

707 708 709

Wendt, C., Frisina, L., & Rothgang, H. (2009, February). Healthcare System Types: A Conceptual Framework for Comparison. Social Policy and Adminstration, 43(1), 70-90. doi:10.1111/j.14679515.2008.00647.x

710 711 712

Wright, D., & Ricketts, T. (2010). The road to efficiency? Re-examining the impact of the primary care physician workforce on health care utilization rates. Social Science & Medicine, 70, 2006-2010. doi:10.1016/j.socscimed.2010.02.043

715

SC

M AN U

TE D

EP

714

AC C

713

RI PT

690 691 692

29

ACCEPTED MANUSCRIPT

716

Tables and Figures

717 Table 1: Countries Selected by Welfare System Type

Liberal Australia Canada New Zealand United Kingdom United States

Corporatist Austria France Germany Italy

Social Democratic Denmark Finland Norway Sweden

SC

719

RI PT

718

AC C

EP

TE D

M AN U

720

30

ACCEPTED MANUSCRIPT

Table 2: Outcomes of Interest for P4P Programs in All Countries, by Welfare System Type Healthcare funder commitment Program Share of Start Date Physician Income from Program 1998 4-7%

PIP

Canada (Ontario)

PCR P4P

2002

10%

New Zealand

PMP

2006

Varies, subject to PHO discretion.

United Kingdom

QOF

2004

Medical Association Responses

Participation Rates

82%

Favourable. American Medical Association (AMA) involved in program development.

M AN U

Australia

Physician acceptance

Program design Target- or ParticipationBased

Program Effects

Target-based.

Impact not significant. Limited influence on behaviour due to modest payments, administrative burden of claiming incentives. Modest improvement in most services.

Cashin and Chi, 2014; Greene, 2013; Kecmanovic and Hall, 2015.

RI PT

P4P

SC

Country

Favourable. Ontario Medical Association (OMA) negotiated program with Ontario government. Unenthusiastic. Physician representative involved in negotiations, but program seen as too unrewarding. Favourable. British Medical Association (BMA) negotiated program with UK

TE D

Typology

AC C 20%

31

Sources

4-40%

Target-based.

Almost 100%

Target-based.

Modest improvements in quality of care.

Buetow, 2008; Cashin, 2011.

Almost 100%

Target-based.

Positive impact on quality of care. Reduction of disparities in risk factor control, especially in lowest-

Cashin, 2014; Doran, et al. 2011; Gravelle, et al. 2010; Latham and Marshall,

EP

Liberal

721

Anderson et al., 2006; Li et al., 2014.

ACCEPTED MANUSCRIPT

Country

P4P

Healthcare funder commitment Program Share of Start Date Physician Income from Program

Physician acceptance Medical Association Responses

Program design Target- or ParticipationBased

Participation Rates

M AN U

SC

government.

2001

Denmark

n/a

n/a

Finland

n/a

n/a

2%

Kristensen, Bech, and Lauridsen, 2013; Olejaz et al., 2012. Vuorenkoski, Mladovsky, and

Target-based.

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

AC C

Social Democratic

2015; Lee et al., 2011.

Est. 70%

32

Sources

performing practices. Many physicians already completing guidelinerecommended care. GPs may be overcompensated. Indication of unintended consequences such as cheating and performance decrease in nontargeted indicators Positive impact on quality of care. Perceived as costeffective. Combination of payers and complex payment rules reduce GP responsiveness to incentives.

Favourable in principle. Some concerns with program administration and incentive sizes. Concerns about accuracy of measures and data contributes to resentment. n/a

TE D

IHA

EP

United States (California)

Program Effects

RI PT

Typology

Damberg et al., 2009; McDonald, White, and Marmor, 2009; Rosenthal and Dudley, 2007; Rosenthal, 2014.

ACCEPTED MANUSCRIPT

P4P

Healthcare funder commitment Program Share of Start Date Physician Income from Program

Physician acceptance Medical Association Responses

Participation Rates

n/a

n/a

n/a

n/a

n/a

Sweden

County P4P

2010

2-4%

Responses to specific programs unknown, but growing acceptance of performance paradigm.

Unknown. Programs are localized.

Austria

DMP

2011

Unknown.

France

CAPI / ROSP

2009

5-7%

Favourable. Program developed in partnership with GP representatives and sickness funds. Strongly opposed at first, but moderated position to accommodate minority support for program.

Germany

DMP

2002

33

Sources

Mossialos, 2008. Ringard et al., 2013. Anell and Glenngård, 2012.

n/a

Target-based.

Knowledge of program effects limited. Issues with validity of clinical indicators.

15%

Participationbased.

Knowledge of program effects limited. Still in implementation stage as of 2011.

Schmidt, Tarver, and Ruppe, 2012.

40% by 2011. Enrolment automatic following ROSP reform (3% opted out).

Target-based

Marginal impact on quality of care. Minimal impact on length of patient consultations. GPs who enrolled may have met targets regardless.

Bousquet, Bisiaux, and Chi, 2014; Or, 2010; Saint-Lary and Sicsic, 2015; SaintLary et al., 2013.

GP participation varies by condition. National data not available.

Participationbased.

Effective in ensuring continuity of care for chronically ill patients. Modest improvement in

Busse, 2004; Chi, 2014; van der Heuvel et al., 2010.

M AN U

TE D

EP AC C

Strongly opposed at first, moderated position when program

Program Effects

n/a

SC

Norway

Unknown.

Program design Target- or ParticipationBased

RI PT

Country

Corporatist

Typology

ACCEPTED MANUSCRIPT

722

2003

Performance: 0.4% Participation: 5.4% Compliance: 0.1%

Participation Rates

became firmly established. Generally mixed opinions on mixing pay and performance – concerns about threat to autonomy, too much influence from sickness funds, pharmaceutical industry. Favourable, as programs were successfully negotiated with GP associations.

Can be as high as 70% regionally.

34

Program Effects

Sources

health outcomes.

SC

Medical Association Responses

Program design Target- or ParticipationBased

RI PT

Physician acceptance

M AN U

EmiliaRomagna P4P

Healthcare funder commitment Program Share of Start Date Physician Income from Program

TE D

Italy (EmiliaRomagna)

P4P

EP

Country

AC C

Typology

Performance: 26.0% Participation: 98.7% Compliance: 14.6%

Mix with dominantly participationbased.

Significant but modest association between improved health outcomes and size of financial transfers (excepting the compliance program).

Florentini et al., 2011; Lo Scalzo et al., 2009.

ACCEPTED MANUSCRIPT

Figure 1: Welfare Systems Influence on P4P: Intersection of Decommodification and Social Stratification

M AN U

SC

RI PT

723

724

Figure 2: Conceptual Framework of the Effects of Welfare Systems on P4P Experiences

AC C

EP

726

TE D

725

727 35

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Acknowledgement: We would like to thank Marc-André Gagnon and Leslie Pal for their comments on earlier versions of this manuscript. Rianne Mahon provided useful guidance on the welfare state literature. We thank the two anonymous reviewers for their insightful comments that helped us to improve the quality of the manuscript. Usual disclaimer applies.

ACCEPTED MANUSCRIPT

Research highlights: This study bridges the social policy and physician payment literature

•

Welfare systems explain variation in experiences with pay-for-performance

•

Liberal systems make enthusiastic use of pay-for-performance (P4P)

•

Social democratic systems have minimal interest in P4P for general practitioners

•

Opposition in corporatist systems results in participation-based P4P

AC C

EP

TE D

M AN U

SC

RI PT

•