Patients’ expectations of variation in quality of care relates to their search for comparative performance information

The public release of comparative performance information (CPI) is common in many countries [1]. The 2006 Healthcare Market Regulation Act led to better availability of CPI in the Netherlands. Choice for consumers and providers became a cornerstone of this new healthcare system based on market elements and competition. CPI can include information about service, patient experiences, and quality indicators for clinical care (structure, process, and patient outcomes). The CPI in the Netherlands includes items related to the hospital (e.g. ranking of 100 hospitals) and to condition-related factors (e.g. patient experiences, waiting lists, and annual patient volumes). The purpose of making this information publicly available is to enable healthcare consumers to choose high-quality healthcare [2] and to empower them to make an informed choice about healthcare [3].

There is no firm evidence that CPI influences patient choices [4]. American and British studies have shown that the actual use of CPI for hospital care is restricted to 4–14% of the consumers [5],[6], while the idea of choice appeals to most consumers [7]. Several Dutch studies of CPI for total hip or knee replacements have been performed [8]-[11]. The patients in Moser and colleagues’ study considered CPI to be an additional source of information when they were preparing for a doctor’s appointment. They benefited from the information most when they had to undergo a total hip or knee replacement for the first time [9]. American patients who report a lack of hospital choice for total hip or knee replacement are more likely to be dissatisfied with their surgery [8]. This observation suggests that encouraging patients to engage in provider selection based on quality of care would improve their satisfaction.

The low level of CPI search behaviour for selecting a hospital is partly due to the consumer’s previous experience with a specific hospital [12]-[14] and the unfamiliarity of using CPI for hospital selection [15],[16], as well as the role of the referring physician. Dutch research shows that many patients prefer their primary care physician to be involved in the choice of hospital so they can either take the physician’s advice or delegate the decision [17]-[19]. Furthermore, many consumers are unaware of the opportunity to consult CPI [6],[20],[21].

To raise the level of CPI awareness (an important step in a causal chain towards using CPI) an attendant motive is required so that consumers feel a need for this information [21]. One consumer choice model takes awareness as a starting point [18]. Consumers must be aware that there is CPI, and that it is possible to make a choice. We presume that patients’ expectations are an important pre-step and have a subjective influence on awareness. Studies show that, when consumers start searching for information, the questions and expectations already in their minds will drive the direction of their search [22]-[24]. Having expectations about the practice variation of hospitals and therefore perceiving a risk of receiving poor care might be a motive for using CPI.

Despite on-going efforts and investments in the collection, production, and dissemination of CPI for the public [15], there is no firm evidence that CPI influences patient choices [4]. In an attempt to bridge this gap, we hypothesise that consumers who expect to find quality differences between hospitals are those who search for CPI. We tested whether patients’ expectations of variation in quality of care affect their reported search behaviour for hospital performance information, then we adjusted for potential confounders. Furthermore, respondents were asked in what ways they expected hospitals to differ. We also asked them about the most important reasons why they did not search for CPI. We obtained data from patients who had recently been admitted to hospital for an elective total hip or knee replacement. Admission to a hospital for elective surgery can be planned in advance, which gives patients time to search for and look into CPI. These replacement procedures are provided at all 87 hospitals in the Netherlands.

In a cross-sectional study of three types of hospitals, we used a consecutive sampling strategy to recruit 475 patients undergoing a total hip or knee replacement, and we invited them to participate in a paper-based survey. To make our sample representative, we included patients from a university hospital, a teaching hospital, and a community hospital. We included adult patients undergoing a primary hip or knee replacement because we expected that previous surgical experience (as in the case of a secondary replacement) would bias the selection. The annual patient volumes were 53 for primary hip replacements and 64 for knee replacements in the university hospital; 232 and 170, respectively, in the teaching hospital; and 236 and 153, respectively, in the community hospital. Data collection took from 5 to 9 months (September 2009 through July 2010). The nurse or anaesthetist who prepared the patient for surgery personally gave the survey questionnaire to the patient at the preoperative appointment 4 weeks prior to the operation. Reminders were sent 2–3 weeks later in the teaching and community hospitals, but no reminders were sent in the university hospital because permission for this was not granted.

The institutional ethics committee reviewed the study protocol in accordance with local regulations in the Netherlands, and they concluded that the study was not subject to the Dutch Medical Research Involving Human Subjects Act.

The primary outcome measure was the self-reported search for CPI, i.e. ‘Did you search for additional CPI to compare hospitals after it became clear that you needed surgery?’ (answer: yes or no). Respondents did not search for CPI were asked to select their motive from a list of seven pre-listed reasons with an option to add one. The reasons were based on a literature search (unpublished search) and focus group interviews with consumers [25]. Other measures concerned the previous treatment in the selected hospital: consumers’ perception of being well-informed to make a decision, the general practitioner’s (GP’s) role in advising a choice of hospital (all dichotomous variables), and expectations of variation in quality of care between hospitals (large, small, or no difference in quality of care). If respondents expected quality-of-care differences, they were asked to specify these differences for 14 pre-listed factors that were available from the Dutch internet sites KiesBeter.nl (‘Choose Better’) and Independer.nl. Both sites based the factors they listed on a set of quality indicators that the Health Care Transparency Programme listed in 2009 [26]. Focus group interviews with consumers [25] and the annual list of a Dutch magazine, Elsevier [27], also presented factors to be considered. The factors included the available CPI for hip and knee replacements in the Netherlands, which gave the name of the organisation and the clinical performance of the care providers. Formally, ‘distance’ and ‘reputation’ are not CPI, but they were included because they are important to patients [9],[11],[28]. In this sense, such information can be seen as part of the performance of the healthcare system. We used a broad definition of performance, as did Van Loon and Tolboom, who defined three information types: (1) factual information (names, addresses, and type of provider), (2) quality information based on performance, and (3) quality information based on consumer experience [29]. The questionnaire contained items about demographic variables (age, education, and type of replacement [hip or knee]), and hospital characteristics (type and patient volumes for hip and knee replacements).

We used descriptive statistics and frequency tables to describe our study population’s demographic variables (age, gender, and education), previous treatment in the current hospital, awareness that hospital performance information is available for comparison, receipt of choice options from GPs, the search for CPI, and expectations of quality differences in hospital care (Table 1). To address the issue of representativeness, we compared the characteristics of the participants in our study with the characteristics of a larger sample of 1508 Dutch patients who underwent a hip or knee replacement [30].

We included data only for those respondents who provided valid answers for the core items of our study: search for CPI and expectations of quality differences in hospital care. The respondents were dichotomised into a group 65 years or younger, and a group older than 65 years. The education variable (the level of education) was measured on a four-point scale (none, low, middle, and high), and it was dichotomised for analytical purposes into low and high levels of education.

We used logistic regression to analyse the relationship between expectations of quality differences (independent variable) and searching for CPI (dependent variable) for patients who underwent a hip or knee replacement (Table 2, model 1). In order to correct for possible confounders, we also performed this univariate analysis for the demographic variables (e.g. age, gender, education, and type of replacement), hospital type, GPs’ role in advising a choice of hospital, awareness of available information for comparing hospital performance, and previous treatment at the current hospital. Potential confounders (p <0.2) in univariate analyses were added to the multivariate model, and we examined their effects on the beta coefficients. Any variables resulting in a change in the beta coefficient of more than 10% were included in the final model. We compared the univariate analysis (model 1) with the multivariate analysis (model 2) for odds ratios (ORs) with 95% confidence intervals (95% CIs). An association was considered statistically significant for p <0.05.

We present descriptive statistics for the factors that we expected to differ between hospitals, as well as statistics for the reasons for not searching CPI. In describing these reasons, we distinguished between respondents who expected differences in the quality of hospitals and those who did not. We used SPSS 18.0 for all analyses.

CPI makes the variation in quality of care between hospitals transparent. This study shows that the number of people who report having searched for CPI is still limited, but may increase if patients become more aware of the quality-of-care variation of hospitals. However, this will be difficult to achieve because people who feel no need for more information – e.g. based on a lack of expected differences in quality of care – do not search for CPI. Awareness as a prerequisite for the use of CPI should not be limited to having knowledge about the existence of CPI and where to find it; awareness should also extend to the quality-of-care variation of hospitals.