Responsibility for choice of hospital
In the present study the GPs reported that they chose the hospital on behalf of 76% of patients. This result appears to contradict results from national Danish surveys of patients’ experience with hospitals: in 2004 46% of in-patients treated in the study area reported, that they chose the hospital; among elective in-patients 89% were aware before being hospitalised that they were free to choose, and 52% of these patients chose the hospital by themselves [26]. The divergent findings may be interpreted as an indicator of shared decision making. When 30% of patients reported that their GP’s recommendation influenced their own choice of hospital [26], the GPs may have perceived that they chose the hospital on behalf of the patient.
The results indicate that patients choose the hospital to a lesser degree than policy makers (politicians and administrators) want them to do to improve management of the public health care sector by introducing a proxy for the market mechanism. Other studies have found that GPs appear to question whether choice is valuable to patients [29], and whether patients really want to choose the hospital [4]. GPs’ choice behaviour varies by GP [29] and by the patients’ diagnoses [39], English GPs being more likely to offer choice to patients, who are in need of a routine intervention, elective patients, and patients who are relatively healthy [4].
One study distinguished between ‘choice enthusiasts’, ‘choice sceptics’ and ‘choice paternalists’ [29]. The present study did not enable us to divide GPs into such subgroups, but confirming results from other studies [4, 30] several GPs expressed reluctance to provide advice to patients, because they did not consider this task a part of their job [30, 40]; considered this task too time-consuming for a consultation [4], or distrusted data published by the providers [29] and wanted to forestall blame for presenting faulty data [23]. This behaviour may reflect an attempt to minimize the length of each visit to the GP. However, at a more general level GPs’ behaviour may reflect a ‘logic of care’ rather than a ‘logic of choice’ [41] - GPs making choices based on their professional views on patients’ needs and wants, rather than as agents acting in a market place enabling patients to make informed choices in line with the neoclassical standard model.
Factors determining GPs’ choice of hospital on behalf of patients
Short distance to hospital was the most important factor behind GPs’ choice of hospital. Numerous other studies of GPs’ actual referral pattern and patients’ choices in structurally different health care systems likewise indicate that short distance strongly influences patients’ and GPs’ choice of hospital [26, 33, 42]. Studies of GPs’ hypothetical referrals and patients’ hypothetical choices have led to other results with GPs emphasizing the importance of short waiting time and the GP’s impression of quality at the alternative departments [35, 43], while patients facing a hypothetical choice emphasized the importance of data on structure quality and attributed little weight to waiting time [44].
Different findings in studies of GPs’ and patients’ choice behaviour may reflect differences between how GPs and patients think they ought to choose the hospital and how they actually make the choice, one study finding significant differences between GPs’ response to hypothetical case stories and their actual referral pattern [45]. Another reason could be international institutional differences with regard to subsidization of transport costs and the length of waiting times.
The small influence of waiting time on choice may be considered to be remarkable, as the media and politicians at the national level consistently focus on waiting times as a major performance measure and challenge, but other studies of choice of hospital likewise found only a small influence of waiting time on choice. Cataract patients generally accepted waiting times of three months and less, while waiting times of six months or more were perceived as too long [46, 47]. In a hypothetical study patients reported that for each additional hour of travel time they would, on average, require a reduction of in the waiting time of 2.3 months [48]. The results of these studies and the present study may partly explain why differences between waiting times at hospitals persisted more than a decade after the introduction of free choice of hospital, but they may also reflect, that a minority of patients are treated as elective patients.
In the present study we focused on the influence of GPs’ sources of information about departments/hospitals and factors commonly found to influence the GPs’ choices. However, GPs’ choices on behalf of patients may be influenced by other agendas independent of the individual patient, i.e. GPs may refer patients to a local hospital to contribute to its continuing existence [30].
GPs’ use of sources of information on quality and service
In the present study GPs were less likely to use official information on quality and waiting time than proxy-measures from informal sources like their own and other GPs’ and patients’ experience with regard to quality and waiting time. This result was consistent with other studies of GPs’ or patients’ utilisation of sources of information, which have found very little utilisation of such sources [49] and refer to GPs as having “a sort of ‘mental filing cabinet’ of informal information or soft intelligence”[28]. The GPs’ experience with cooperation with various departments or hospitals was very important for the GPs’ choice. GPs’ responses indicated that their experience with specific departments was the most important factor, but many GPs attributed their choice to their experience with a hospital in general rather than the individual department, thereby indicating that they generalised their experience from one or more departments at a hospital to other departments at the hospital as a whole – a kind of ‘halo’-effect.
The strong influence of informal data sources like patients’ previous experience on choice and advice on choice may reflect lack of official information on quality or waiting time or that GPs are suspicious of published data on performance, viewing such data as “spin” [29]. Several respondents commented that use of web-based information was too time-consuming compared to data on paper; their memory of previous referrals, and asking the patient to call one or more hospitals or the county’s patient’s advisors for information.
Some GPs wrote that they intended to use web-based information more in the future. Such statements may reflect expectations that more experience and improved IT will ease their access to the web or lack of experience. When the present study was performed approx. 86% of Danish general practices had access to the internet, and a little less than half of the practices used the access each day [50].
Implications
Further research is warranted on the interaction between GP and patient in choice of hospital, preferably by direct observation of the referral process followed by interviews with the GP as well as the patient about their views on the referral process including their experience of responsibility for the choice.
The findings in the present study support results from studies of patients’ choice behaviour which indicate that patients and their agents do not act as the autonomous customers assumed in market-resembling models for management of the public sector. When agents act on patients’ behalf they tend to utilise informal sources of information – even when systematically collected and published information on service is available. One implication of the major influence of previous experience with hospital departments may be a tendency to inertia in referral patterns.
Limitations of the study
The response rate in the present study was 52%, which appears to be quite normal for studies performed in general practice.
The choice of study method meant that we did not observe the process of choice, and only reasons we were aware of beforehand were included in the study, but the questionnaire was validated, and the respondents were offered the opportunity to comment on the reasons and did not refer to reasons not mentioned in the questionnaire.
The respondents could report any number of reasons and we did not ask them to quantify the importance of each reason, because this would complicate the data collection and probably reduce the response rate. We assumed that the cumulative importance of a reason for choice of hospital was proportional to the frequency it was quoted by the GPs, but this may not necessarily be the case: a comparison of two Dutch studies published recently may indicate that frequency of reporting may give results which differ from estimations of importance by way of a choice experiment [25, 44].
Respondents did not differ from non-respondents with regard to age, gender or county, but GPs with a stronger than average interest in subjects concerning choice of hospital may be especially likely to participate in the study. Therefore the study may exaggerate the impact of each individual factor on choice of hospital.
Usually studies should be performed prospectively to reduce bias, but in the present study we chose a retrospective design in order not to influence the GPs’ choice behaviour. Our choice of design increased the risk of recall bias, and the GPs may have reported factors which they thought ought to have influenced their choices rather than the decisive factors. For example GPs may have hesitated to quote media reports as an important source of information. GPs probably are very conscious about their use of some sources of information like websites, while the importance of some sources may be underestimated, because their utilisation is more nebulous, like feedback from patients or media reports. Presumably patient characteristics influenced the GPs’ choices but not their willingness to participate. Therefore patient characteristics presumably did not introduce bias in the study.
The study included a large number of statistical tests. Some of the statistically significant associations in univariate analysis may be due to mass significance rather than causality.
The study was performed thirteen years after the introduction of free choice of public hospital within the study area and eleven years after the introduction of free choice of public hospital at the national level. Patients’ awareness of their right to choose was high. Therefore, even though the study was performed at a specific time in the process of introducing free choice of hospital, we find it most likely that studies performed a few years before or after the present study would not have led to results which were very different from those of the present study.