Discussion

Comparative information about patients’ experience with providers collected through survey tools are becoming increasingly more common and used in the assessment of providers by individuals and payers (Smith et al., Reference Smith, Mossialos, Papanicolas and Leatherman2010). Such information about provider performance is often used to evaluate the overall performance and to identify differences between, for example, public and private providers without critically reflecting about underlying factors. The results in this study indicate that factors that are of relevance for how well providers perform according to such surveys are more or less possible to control for providers. This might in turn have consequences for the use of such information by individuals and payers. Moreover, having in mind the potentially conflicting objectives of health care is important when collecting, analyzing and acting upon comparative information about providers (Browne et al., Reference Browne, Roseman, Shaller and Edgman-Levitan2010). Primary care should fulfill a number of conflicting objectives and there is no simple solution on how to best organize primary care services to achieve fulfillment of one objective without adverse consequences for the fulfillment of other objectives (Lamarche et al., Reference Lamarche, Beaulieu, Pineault, Contandriopoulos, Denis and Haggerty2003; Bevan et al., Reference Bevan, Helderman and Wilsford2010).

The fulfillment of objectives related to patient satisfaction was analyzed with respect to several characteristics of providers, based on information from a national patient survey in primary care. The sample of slightly more than 400 primary care providers located in three county councils represents about one-third of all primary care providers in Sweden. The principle for payment and conditions for accreditation of providers as well as for registration of individuals with a provider in the three county councils are similar to the primary care models practiced in a majority of the county councils in Sweden. The availability of alternative providers is, however, higher in the sample compared with in many other county councils.

On an overall level, the finding that the perceptions regarding the different questions were correlated implies that providers that have satisfied patients regarding one question also tended to have satisfied patients regarding other questions. The models used to analyze the six chosen questions indicate that there were systematic variation in patient satisfaction with respect to both organizational and structural factors, including the mix of registered individuals among providers. In line with the finding about correlation between the questions, different factors tended to affect the different aspects of patient satisfaction in similar directions although to different extent. The factors reflecting organizational and structural differences included in this study are, however, not the only factors of relevance for patients’ perceptions about the quality of services provided among primary care providers. Factors reflecting expectations and preferences in primary care among individual patients are for instance not included in the models.

Differences in patient satisfaction with respect to type of owner were found for two questions. Private providers were associated with different perceptions regarding patient satisfaction for one aspect, that is, overall satisfaction. For all other aspects, no significant correlation was found between patient satisfaction and the private/public dimension. In another recent Swedish study, based on a population survey in the same county councils as included in the present study, it was found that a majority of the respondents had no opinion about whether private health care providers provide better services than public providers (Glenngård et al., Reference Glenngård, Anell and Beckman2011). Moreover, regarding continuity of care for doctor visits, independent private providers were associated with higher patient satisfaction.

With regard to location, providers in the two largest county councils (Region Skåne and VGR) were associated with a lower continuity of care compared with the county council of Halland. Being located in the two biggest cities in the sample (Malmö and Göteborg) was associated with a lower satisfaction among patients regarding several of the included dimensions of patient satisfaction. One possible explanation is that individuals who live in large cities have higher expectations on primary care services in general and on accessibility in particular due to the short distance to alternative health care providers, including emergency care hospitals.

The mix of registered individuals was found to be correlated to patient satisfaction for all aspects investigated. The most common factor associated with a lower satisfaction among patients was high social deprivation (CNI) among registered individuals with a provider. All aspects of patient satisfaction included in this study were negatively correlated to this parameter. This is in line with findings by Kontopantelis et al. (Reference Kontopantelis, Roland and Reeves2010) and Campbell et al. (Reference Campbell, Ramsay and Green2001b) who have demonstrated that patients living in socio-economically deprived areas tend to be less satisfied with services. The opposite was found for overall illness of individuals (ACG). A high overall illness among registered individuals was found to be positively correlated with patient satisfaction for almost all aspects included in the study. This may, however, also reflect the fact that the elderly are more satisfied since the number of diagnoses and the number of health care visits commonly increases with age. A limitation in the current study is that no variable reflecting only age of registered individuals was included.

There was no systematic variation in patient satisfaction with respect to a high total number of visits per registered individual. But all aspects of patient satisfaction investigated were positively correlated with a high proportion of all visits being with a doctor. Thus, the organization of work seems to be an important factor for patient satisfaction. Possible explanations to this finding are related to both expectations among patients and the character of the visit. If patients expect that they will see a doctor there is a risk of being dissatisfied with the visit if they get to see a nurse instead, irrespective of the actual content or quality of the visit. Moreover, if an increased proportion of all visits occur with nurses, the content of nurse visits will change since more focus has to be directed at the clinical content of the visit and thereby perhaps less focus is directed at, for example, responsiveness of staff in relation to their patients. One of the ideas of the model of team-based primary health care, which is practiced in Sweden, is, however, that the substitution of physician labor input for non-physician labor input, such as nurses, has benefits for patients and also leads to efficient use of resources. It is expected to shorten waiting times and improve services as physicians can devote more time to patients with severe problems (Jessup, Reference Jessup2007; Yarnall et al., Reference Yarnall, Østbye, Krause, Pollak, Gradison and Michener2009). Also the primary care models practiced in the three county councils included in this study are based on the rationale that individuals should register with a primary care unit rather than with a general practitioner (Anell, Reference Anell2011). In light of the shortage of physicians in Swedish primary care, this is indeed important in order to improve access to primary care services. However, the results in this study indicate that satisfying patients’ perceptions about quality of care while at the same time using scarce resources for primary care in an efficient way is a difficult task.

Size, more specifically a smaller number of registered individuals were associated with a higher satisfaction among primary care patients for all aspects. This was especially pronounced with regard to the continuity of care for nurse visits. This finding is in line with previous studies, where it has been demonstrated that smaller practices are associated with a higher patient satisfaction regarding access and continuity of care (e.g. Campbell et al., Reference Campbell, Ramsay and Green2001a; Kontopantelis et al., Reference Kontopantelis, Roland and Reeves2010). Small practices may, however, experience diseconomies of scale in other dimensions. The payment principle in Swedish primary care is primarily based on fixed capitation combined with a comprehensive financial responsibility for services. This gives providers incentives to have many registered individuals in order to balance off costs for patients with great needs with individuals expected to have lesser demands (Barnum et al., Reference Barnum, Kutzin and Saxenian1995; Jegers et al., Reference Jegers, Kesteloot, De Graeve and Gilles2002). One possible way to solve this potential dilemma is to organize several smaller units in a chain of providers in order to balance off costs between providers if necessary.

A higher patient satisfaction associated with a higher overall illness among registered individuals is at least theoretically possible for the providers to control with other means than geographical location in a system based on freedom of choice of provider for citizens. By hiring a specialist in diabetes or geriatric care, for instance, and communicating this, it might be possible to attract individuals with certain diagnoses and thereby a high overall illness. On the other hand, such individuals are potentially costly, and providers may also need economic incentives to attract individuals with expected great care needs, that is, risk-adjusted care payment (Le Grand, Reference Le Grand2009; Miller, Reference Miller2009). More or less sophisticated systems for risk-adjusting the capitation to providers is practiced in the three county councils included in this study. In Halland, a model based on age is practiced, and in Skåne and VGR, a model based on risk-adjusted capitation with respect to both overall illness and social deprivation is practiced. This theoretically gives providers in Skåne and VGR incentives to attract less healthy individuals, whereas the opposite is the case in Halland (Anell, Reference Anell2011).

Providers risk not being favored by individuals when they make a choice of provider if they are located in areas with high socio-economic burden. This risk can to some extent be mitigated by establishing only in areas with a low socio-economic burden. This is of course not a desired scenario as socially deprived individuals have a greater expected need of health care services and thus would benefit from more and better quality services rather than the contrary. The whole idea behind expanded patient choice and provider competition is that it should improve the efficiency, quality and responsiveness of providers in relation to their patients through market mechanisms (Le Grand, Reference Le Grand2007; Anell, Reference Anell2011). As patients can shift provider in case they are not satisfied with their current provider, they can cause a loss of income for the providers (Dixon and Le Grand, Reference Dixon and Le Grand2006; Fotaki et al., Reference Fotaki, Roland, Boyd, McDonald, Scheaff and Smith2008). If providers choose not to establish in areas with a high socio-economic burden, this mechanism risks failing, leading to increased inequalities. People in lower socio-economic groups often have lesser capabilities to exercise choice compared with people with higher income and/or education who are more articulate, have better skills in searching for information and are more able to travel to non-local providers (Barr et al., Reference Barr, Fenton and Blane2008; Fotaki et al., Reference Fotaki, Roland, Boyd, McDonald, Scheaff and Smith2008).

In order for increased patient choice to contribute to the efficiency, quality and responsiveness of health care systems, individuals must actually use the information available to make informed choices (Hibbard and Peters, Reference Hibbard and Peters2003). Previous Swedish studies have shown that individuals are interested in making a choice of provider (Anell et al., Reference Anell, Rosén and Hjortsberg1997; Hjelmgren and Anell, Reference Hjelmgren and Anell2007), but that they are rather passive in their search for information (Nordgren and Åhgren, Reference Nordgren and Åhgren2010; Glenngård et al., Reference Glenngård, Anell and Beckman2011). Individuals tend to get their information from providers that they have previously been in contact with. Hibbard and Peters (Reference Hibbard and Peters2003) suggest three processes to enhance the individuals’ use of comparative information: lowering cognitive effort, helping people to understand the experience of a choice and highlighting the meaning of information. The results in this study perhaps have implications primarily for highlighting the meaning of information. Variation in patients'assessment of provider performance was found with respect to the location, organization of work as well as to the mix of registered individuals. This adds to the complexity of using aggregated comparative information about providers to base individual choice of providers upon. If information about provider behavior is largely influenced by factors that are not related to actual differences in provider behavior, such information is perhaps not accurate to base decisions about choice of provider upon. It is important but perhaps even more difficult for citizens than for other actors to consider underlying factors affecting aggregated information about providers. One solution could be comparing providers in groups with similar conditions, regarding mix of registered individuals instead of comparing all providers without reflecting upon underlying factors of importance for patients’ assessment of provider behavior. Then the risk of not being favored when individuals make their choice of provider associated with being located in socio-economically deprived areas is less prominent. On the other hand, information showing that individuals in socio-economically worse of groups tend to be less satisfied with services provided indicates that providers serving such individuals might have to change their organization of work to better meet the needs of their patients.

The analysis about patient satisfaction in primary care related to the characteristics of individuals in this study was conducted based on aggregated information about mix of individuals with each provider. Different individuals, however, have different needs and expectations regarding primary care and thus prefer different solutions to the delivery of such services (Hjelmgren and Anell, Reference Hjelmgren and Anell2007). Moreover, information about patient satisfaction is likely not the only kind of information that individuals are interested in when making their choice of provider. Clinical performance measures might be equally important. If people have access to and actually use accurate comparative information about providers, they may also choose a provider that is best capable of meeting their individual needs and expectations. This is indeed important for the intended mechanisms with increased patient choice in primary care to work in practice.