Introduction

Intensive care funding has recently been changed in some European countries [1, 2] in order to meet the increasing cost of health care. It is well known that different methods of reimbursement are in use between countries; however, there is presently little information available on how these methods compare with one another and the relative merits of each. We attempted to obtain in-depth information on this subject from a volunteer sample of ESICM members. The aim of this survey was to collect descriptive information on the different methods of reimbursement in use in intensive care units and to evaluate the relationship between financing methods and staff satisfaction.

Methods

The Research Group on Health Economics of the European Society of Intensive Care Medicine (ESICM) set up a Steering Committee in order to create a questionnaire in February 2007. The members of the Steering Committee approved the final survey questionnaire in June 2007. The Research Committee of ESICM evaluated and accepted this study for the Survey of the Month series in July 2007. An e-mail launch was organized by ESICM, in which the 4,275 active members of the ESICM at the time of the survey were asked to participate. The call for participation e-mail was sent out on 3rd September 2007, and then a reminder was launched on 24th October 2007. The website was closed on 31st October 2007.

Survey questionnaire

The questionnaire was anonymous and did not have any ethical issues; however, name of country, city, and intensive care unit (ICU) were collected. The questionnaire consisted of 22 questions divided into two parts: in part I the questions collected general ICU data about financial situation, and in part II the questions focused on ICU reimbursement practices. There were altogether 7 questions in part I and 15 in part II. The questionnaire in pdf format can be found in the Electronic Supplementary Material. The professional role of the respondent was considered important, because the opinion of clinicians may be different from those who play managerial role in ICU. Senior medical staff was defined as fully qualified intensivist, who can work without supervision, as opposed to junior medical staff. ICU/hospital managers, audit/research professionals, and nurses could also participate in the survey. The level of ICU care provided and the position of ICU (university versus not) were also included, these factors being known to have an impact on ICU funding [3]. Difference between level II and level III ICU was defined such that level II can only provide one or two organ support including mechanical ventilation, while level III can provide full organ support. Level I care was categorized as basic ICU care for monitoring only. University hospital was defined as those attached to a medical school or affiliated to a university. The questions about reimbursement practices included items on costing awareness and resources available, and whether ICU reimbursement was irrespective or not of hospital funding. Two questions were focused on identifying factors on which the reimbursement practice of the respondent ICU was based. Further data were collected on practical aspects of ICU funding/reimbursement, aiming to obtain information about timings of reimbursement. Personal satisfaction with the current ICU reimbursement system was finally collected on a Likert scale using the statement: “I am satisfied with our current ICU reimbursement system in use.” Possible answers were: “strongly disagree,” “disagree,” “neither agree nor disagree,” “agree,” and “strongly agree.” The questionnaire also provided an open space to report opinions on the main strengths and limitations of the reimbursement system in force.

Statistical analysis

Answers to the questionnaire were automatically collected in a Microsoft Excel worksheet. Statistical analysis was performed using SPSS version 11.5 for Windows. Unless otherwise specified, categorical variables were expressed as absolute values with percentages. Chi-square test was used for categorical data and Mann–Whitney test for interval data. A p value of <0.05 was considered statistically significant.

Results

There were altogether 449 respondents who completed the web-based questionnaire. Two answers were excluded due to inadequate data entry. Out of the remaining, 348 answers (77.8% of total) were sent from European ICUs, and the others were from 26 other countries, in order of frequency Saudi Arabia (n = 16), India (n = 13), and Australia (n = 12). Within Europe, the UK (n = 55), Italy (n = 50), and Germany (n = 32) had the highest number of respondents (Table 1). The majority of respondents (83.1%) were senior medical staff, and only 6.7% were hospital managers. Almost two-thirds of respondents worked in university hospitals (63.0%) and 7.4% in privately funded hospitals. Level III ICU care was reported by 78% of respondents, and level I by only 0.5%.

Table 1 Number of answers by country (n = 447)
Full size table

Half of the respondents (n = 230, 51.5%) stated that they received detailed financial information about their ICU, the commonest interval for reporting expenditure being monthly (n = 135, 54.9%), and 14 (5.7%) respondents stated that their ICU received expenditure reporting after each case. Only 69 respondents (15.4%) stated that they could identify each cost item for each patient, while the majority could only identify cost items in aggregate for all patients. For most respondents, intensive care unit’s reimbursement system was included in the hospital reimbursement (n = 263, 77.6%); however, 118 out of these respondents (47.4%) were able to identify ICU costing.

ICU reimbursement system was most commonly based on previous year’s ICU expenditure (n = 173, 51.0%); use of diagnosis-related group weights was reported by 122 (36%) respondents (Table 2). There were some other factors taken into account, such as patient day, severity of illness, nursing workload score, etc. (Table 3), but 137 respondents (40.4%) stated that reimbursement had no added elements. These listed factors were used to reimburse full intensive care costs for 167 respondents (53.4%) only, which implies that, for 46.7% of respondents, a proportion of costs were reimbursed in this way. Information on case-by-case reimbursement of ICU stay was reported as available by 114 respondents (33.6%); however, for 145 (42.8%) respondents, reimbursement of hospital stay could be identified. If ICU reimbursement at the end of financial year was lower than actual costs, the hospital and the health insurance fund would cover the costs for 224 respondents’ ICU (66.1%). According to Likert-scale-based personal opinion, 68 respondents (21.1%) agreed with the statement that they were satisfied with their ICU’s reimbursement system. The number of original categories in Likert scale was reduced to ensure statistical power for analysis: “strongly agree” and “agree” answers were handled as “agree,” and similarly “strongly disagree” and “disagree” were unified into “disagree” option. Also, it has been proven that, when only three levels are used, the central tendency bias prevents many responders to take away from the middle. That is why we used five categories in the questionnaire and three in the analysis.

Table 2 List of primary factors taken into account in ICU reimbursement/funding systems (n = 339)
Full size table
Table 3 List of secondary factors taken into account in ICU reimbursement/funding systems (n = 339)
Full size table

We selected and analyzed a homogenous sample from our database; the selection criteria’s were: (1) responses from European ICU, (2) answers by senior medical staff or hospital manager, (3) nonprivate hospital, and (4) level II or III care. This sample consisted of 306 respondents. The satisfaction of respondents (based on Likert scale) and their answers to ICU reimbursement questions were analyzed by chi-square test. We found that those who received detailed financial information were significantly more satisfied (p = 0.010) with their reimbursement system (Table 4). Respondents were significantly less satisfied (p < 0.001) with their system in those units where ICU reimbursement was included in the hospital reimbursement system (Table 5). Regarding ICU funding elements, the most satisfied with their ICU reimbursement system were those respondents from ICUs where nursing workload score was used (p = 0.001) (Table 6). There was no statistically significant difference found in relation to any other funding elements and respondents’ satisfaction.

Table 4 Do you receive detailed financial information about your ICU expenditure?
Full size table
Table 5 Is your ICU reimbursement irrespective of your hospital reimbursement system?
Full size table
Table 6 ICU reimbursement system is based on nursing workload score (e.g., TISS, NEMS)
Full size table

We divided responses from European ICUs into geographic regions: East Europe (n = 44), South Europe (n = 89), West Europe (n = 145), and North Europe (n = 28). We tested regional variance between these geographic regions for all questions; statistically significant differences were found in four answers: (1) bases of reimbursement system, (2) estimation of reimbursement case by case, (3) timescale for receiving reimbursement, and (4) payment for underfunding.

  1. 1.

    As a basis of ICU reimbursement, in North Europe significantly more respondents stated previous year’s ICU expenditure (p < 0.001) (ESM, Fig. 1).

  2. 2.

    Estimation of reimbursement on a case-by-case basis was significantly more common in East Europe (p < 0.001) (ESM, Table 7).

  3. 3.

    Knowledge of reimbursement timing (i.e., <1 month or <6 month) was significantly greater in East Europe (p < 0.001) (ESM, Table 8).

  4. 4.

    In case of ICU underfunding, the health insurance fund covers the cost in ICUs of East and South Europe (p < 0.001) (ESM, Table 9).

Discussion

Intensive care is widely considered as an expensive specialty because of high proportion of staff costs and dependence on up-to-date technology. The official figures from US nationwide data are often quoted, i.e., that ICUs represent 13.3% of hospital costs and 4.2% of national health expenditure [4]. While intensive care may be costly, it may represent good value for money due to the lives it can save. This was confirmed by a recent study from Ridley and Morris [5], who found that, relative to non-intensive-care treatment, the incremental cost per quality-adjusted life year gained (QALY) of treatment in ICU is 7,010 GBP. This figure is much lower than the published QALY values for statin treatments: 10,000–17,500 GBP for secondary prevention and 9,000–119,000 GBP for primary prevention [6]. These data confirm that intensive care is cost-effective; therefore, reimbursement of ICUs should not limit treatment options in any case.

There is no universally “best” funding method available for intensive care units at present. As our results showed, half of respondents stated that their ICU used the previous year’s expenditure for the following year. This form of reimbursement, however, has its limitations, as it does not take into account changes in case mix or medical practices. According to our respondents, this funding method was used in the following European countries: Austria, Belgium, Czech Republic, Denmark, Greece, Italy, The Netherlands, Norway, Portugal, Spain, Sweden, and the UK. The respondents were much less satisfied with their ICU reimbursement system in these countries, with only 16.4% of them agreeing or strongly agreeing on the Likert scale.

We found that diagnosis-related group (DRGs) weights were the second most common form of ICU funding. This was introduced in the USA in the 1960s, initially to evaluate hospital performance. Due to the steady increase in length of stay and healthcare costs by the 1980s, the US Department of Health adopted DRG as the base unit of payment in the Medicare system in 1983 [7]. Studies about the accuracy of DRG funding showed that it induces underfunding [8]. Cooper et al. analyzed the cost and payment for intensive care unit services among Medicare beneficiaries for 1 year (n = 10,657,587) and found that intensive care unit patients cost nearly three times as much as ward patients (US $14,135 versus US $5,571); however, intensive care unit cases were paid at a rate only twice that of ward cases (US $11,704 versus US $5,835). Similar data are available from Europe as well, showing marked differences between countries [9–11].

The Simplified Therapeutic Intervention Scoring System (TISS-28) was introduced in 1996, initially designed to quantify nursing workload. It was some years later that studies were published about the relationship between costs and TISS score. Although Dickie et al. found very good correlation between total TISS score and variable ICU costs (r = 0.93) for a group as a whole, for the individual patient the range of error was up to ±65% of true variable costs [12]. Similar results were published from a pediatric ICU: regression analysis of TISS against ICU costs indicated that TISS was a poor predictor (r 2 = 0.56–0.82) [13]. Despite these data, TISS score is used widely for cost assessment, with the value of one TISS point being found similarly in the range of 35–38 Euros [14, 15]. Our results indicate that respondents from ICUs where some form of nursing workload was used for reimbursement were more satisfied with their funding system.

Severity scoring and ICU costs were analyzed in Germany at the time of DRG introduction. Hoehn et al. investigated the correlation between five scoring systems and German DRG system reimbursement, finding no positive correlation between any score and reimbursement of costs [16]. This explains our finding that severity of illness is only used in 10.6% of ICUs in our sample.

The main strength of our study is that, to our knowledge, there has been no published work so far on comparison of ICU reimbursements between countries. This survey should be considered as a first step towards a prospective study; however, the diversity of our sample has its limitation, which has to be taken into account when our data are interpreted.

The main limitation of our study is that a questionnaire survey is not the best method to assess reimbursements of ICUs across Europe; however, we think that, notwithstanding its limitations, this approach provides some hints that can help improve understanding of this topic. More than one questionnaire may have been sent to different doctors of the same ICU, but we do not consider this as a limitation, since we were interested in the opinion of an average intensivist on how ICU funding was perceived. Double compilation by the same respondent could be excluded from collected data (based on IP address, country, city, and optional e-mail address). The representativeness of our survey may be questioned (447 out of 4,275, 10.4%), but our response rate is much better than achieved in some recent surveys (e.g., data intensive software connectors (DISCO) data). Some comparisons are made based on a small sample, which may imply alpha error; however, in our case, the statistically significant p values are very low, which makes false-positive results unlikely.

Our survey suggests that ICU doctors are more aware of their unit’s financial situation than commonly expected and that they are more satisfied if detailed financial information is given to them. Although an average ICU doctor may not be interested in ICU reimbursement, we would like to point out that cost constraints and its impact on decision making, as part of the new ICU governance, implies knowledge in this issue. The ideal ICU funding system should be irrespective of hospital care; however, defining the best method of reimbursement requires further studies and cannot be answered from our survey. In conclusion, even though the representativeness of our survey can certainly be questioned (447 out of 4,275, 10.4%), nonetheless the results, although far from conclusive, provide some hints that can help to improve understanding of the difficult topic of ICU reimbursement across Europe.