Introduction

In many Western countries, the economic imbalance of public health systems is directly linked to the uninterrupted increase in health care expenses [1]. Clinicians, responsible for the consumption of nearly all of the care and medical goods, play an important role in strategies for directing health care spending [2–4]. Previous studies have shown important cost savings through the application of rationalized prescriptions while maintaining equivalent heath care management [5–11]. In any health system, a medical cost-control strategy involves the optimization of health expenses through the promotion of medical care quality and the application of established care practices. In France, for example, replacing a global endowment health management system with an activity-based financing system rapidly compelled physicians to accept greater accountability in the cost-control problem [12]. Medical prescriptions are therefore meant to be more considered and more reasonable. More and more doctors feel that costs are an important consideration in the medical thought process that leads to prescribing decisions [13, 14].

Nowadays, incorporating cost-consciousness into our daily practice is unavoidable. Intensive care units (ICUs) represent a large portion of health care expenditures, estimated to reach up to 20 % of some hospital budgets [15]. Consequently, reducing costs in these units has become a priority [16]. Changing physicians’ attitudes towards cost control requires preliminary knowledge of prescription costs. Previous surveys, conducted in North America and Europe, have shown that doctors have a poor understanding of the costs of drugs, laboratory tests, and imaging modalities [17–23]. While several studies were aimed at optimizing ICU prescription strategies, only a few small studies performed over ten years ago investigated intensivists’ cost awareness [17, 19, 20].

The aim of the present work was to assess current intensivists’ knowledge of prescriptions costs in a national ICU study and to identify factors influencing the accuracy of cost estimations.

Materials and methods

Study design

A written questionnaire study was performed from May to December 2010. In each participating unit, every junior (residents and medical students) and senior (MD degree) intensive care physician was surveyed by a local correspondent. The characteristics of the participants (age, sex, level of training, financial training) and descriptions of the centers (medical/surgical ICU, academic/nonacademic hospital) were collected. Surveyed prescribers were anonymously asked to individually estimate the true costs of selected prescriptions.

Questionnaire

The questionnaire listed 46 prescriptions (medications and investigations) commonly used for diagnosis and treatment in ICU practice. These were gathered into four groups: drugs, blood products and derivatives, imaging modalities, and laboratory tests (Table 1). For each group, items were distributed in homogenous subgroups defined a priori (Table 1). The selected prescriptions were either the most frequent and/or expensive ones (annual amount) or regarded as essential to ICU practice. The total cost of the prescriptions was 25,595 €. True hospital costs were obtained using the average costs of drugs and blood products and derivatives in the Hospices Civils de Lyon (i.e., the University Teaching Hospital of Lyon, France), while the costs of the imaging modalities and laboratory tests were based on the French national averages. By the end of the study, a questionnaire with correct estimates was sent to each participant.

Table 1 True costs, estimated costs, and typical clinical cases
Full size table

Typical clinical cases

To appreciate the value of cost awareness in the real world, we took into account two clinical situations observed daily in ICU practice: (1) septic shock due to community-acquired pneumonia (case 1); (2) hemorrhagic shock occurring under vitamin K antagonist (case 2). As reported in Table 1, we considered a 7-day ICU management associated with a number of prescriptions. The true costs, amounting in total to 2223 € for case 1 and 7238 € for case 2, were compared to estimated costs.

Statistical analysis

Data were expressed as number (percentage) and as mean ± standard error of the mean (SEM), as appropriate. The data analysis was performed as follows: calculation of response rates, description of physicians’ characteristics, evaluation of the accuracy of the estimates within margins of error defined a priori (±10, ±25, ±50, and >50 %), comparison of estimate deviations (in real and absolute values), and identification of factors influencing cost estimation accuracy. Accuracy was defined by estimates within 50 % of the true cost.

Univariate comparisons were performed using an analysis of variance (ANOVA) for continuous variables, and a Chi-squared test for categorical variables, as appropriate. The independent contribution of physicians’ characteristics to incorrect estimations was tested by a logistic regression analysis. All variables with a p value less than 0.10 following univariate analysis were introduced into the model. Odds ratios (OR) were estimated with a 95 % confidence interval (95 % CI). Statistical analysis was performed using MedCalc® 7.4.3.0 software (Medcalc, Mariakerke, Belgium). A p value of less than 0.05 was considered as significant.

Results

The response rate among the physicians of the 99 participating services was 83 %: 1092 questionnaires were completed from the 1315 surveys handed out. There was no significant difference between the response rates from academic (83 ± 21 %) and nonacademic (84 ± 21 %) hospitals (p = ns).

The characteristics of the respondents are summarized in Table 2. The majority of physicians were under 40 years old (79 %), male (sex ratio 1.5), and operating in medical or medical and surgical ICUs (83 %). As expected, the ratio junior/senior physicians was significantly higher (p < 0.01) in academic hospitals (3.7) when compared to nonacademic hospitals (0.3).

Table 2 Physician characteristics and factors influencing costs estimations
Full size table

Concerning cost accuracy, most estimates were not within 50 % of the true cost for any prescription group (Table 3). Only 315 physicians (29 %) accurately estimated costs within 50 % of the true cost for the total amount (25,595 €). Response errors included an underestimation of 14,756 ± 301 €, i.e., −58 ± 1 % of the total sum. Absolute value deviations were 79 ± 1 % for drugs, 81 ± 2 % for blood products and derivatives, 73 ± 2 % for imaging modalities, and 73 ± 1 % for laboratory tests. As shown in Fig. 1a, drug costs were the most significantly (p < 0.001) underestimated (−64 ± 1 %), when compared to blood products and derivatives (−57 ± 2 %) or laboratory tests prescriptions (−36 ± 1 %). Imaging modality prescriptions were the only costs that were overestimated (7 ± 3 %). As shown in Fig. 1b, most prescription subgroups were underestimated. A clear trend in the overestimation of cheap prescriptions and the underestimation of expensive ones was observed. This was particularly true in drugs estimations (Table 4). For example, the drug subgroup “less than 10 €” was the only one commonly overestimated (961 ± 45 %). In contrast, “more than 1000 €” was the most underestimated drug subgroup (−67 ± 1 %), representing a considerable economic impact (−10,235 ± 196 € for a true cost of 15,358 € for this subgroup of prescriptions).

Table 3 Accuracy of costs estimations
Full size table
Fig. 1
figure 1

a Cost estimations according to level of training of the physicians. For most of the underestimated groups of prescriptions and for the total amount, estimations by senior grade physicians (green bar) were significantly more accurate than those of junior physicians (blue bar). *p < 0.05 versus “seniors”. b Cost estimations according to prescription subgroups. Correct estimations (green bar) were defined as being within 50 % of the true cost, overestimations (black bar) >50 % of the true cost, and underestimations (red bar) <−50 % of the true cost. The “<10 €” drugs subgroup (i.e., the cheapest one) was the only subgroup overestimated by more than 50 % of responders. The more expensive the other subgroups were, the more underestimated they were. 1 <10 € drugs, 2 10–100 € drugs, 3 100–1000 € drugs, 4 more than 1000 € drugs, 5 plasma, 6 red cells, 7 platelets, 8 blood derivatives, 9 basic radiology, 10 echo-Doppler, 11 computed tomography/magnetic resonance imaging, 12 specialized radiology, 13 hematology, 14 biochemistry, 15 toxicology, 16 microbiology

Full size image
Table 4 Average drug estimations
Full size table

Meaningful underestimations were also found in the two considered clinical situations. Using a ±50 % margin of error, our analysis indicated that 393 physicians (36 %) inaccurately estimated costs of prescriptions for case 1 (septic shock), 513 (47 %) for case 2 (hemorrhagic shock). Response errors of physicians averaged −173 ± 46 €, i.e., −8 ± 2 % of the true cost, for case 1, and −2423 ± 102 €, i.e., −33 ± 1 %, for case 2.

For the underestimated groups of prescriptions and for the total amount, the cost estimations of senior grade physicians were more accurate (p < 0.05) than those of the juniors (Fig. 1a). Age, sex, level of experience, hospital characteristics, and financial training significantly influenced the accuracy of cost estimations (Table 2). In multivariate analysis, junior physicians (OR, 2.1; 95 % CI, 1.43–3.08; p = 0.0002) and female gender (OR, 1.4; 95 % CI, 1.04–1.89; p = 0.02) were the only variables independently associated with incorrect cost estimations.

Discussion

The present study shows that, on a national level, intensivists have poor awareness of ICU costs. This knowledge deficit, particularly apparent among junior physicians, is dramatically illustrated in the lack of appreciation of the costs of the most expensive prescriptions.

The burden of the economic situation in health care demands the application of a medical cost-control strategy, urging physicians to provide cost-effective management without compromising quality of care. The goal of a tight cost-control management is obviously not to reduce the level of care but to optimize resources allocated for health, which are not unlimited. Respecting evidence-based medicine, physicians must now make choices when prescribing in order to give cost-effectiveness and optimal care quality [24, 25]. Because of the large portion of health care expenditure directly attributable to the ICUs, this urgent issue is particularly important in critical care medicine [26, 27]. Indeed, on a daily basis, intensivists are faced with new diagnostic tests, specialized disposables, or expensive drugs, which represent a significant part of the growing expenditures of health care [28, 29]. Individually, ICU prescribers play a key role in the critical care cost-containment problem: their medical responsibility is especially linked to the economic impact of the care they provide. Making prescribers responsible requires in-depth changes in prescribing patterns and in the physician’s attitudes towards cost awareness [3, 13]. Being aware of prescription costs is surely the first step in incorporating cost-consciousness into medical prescribing decisions [30–32].

Here, we carried out the largest study to date concerning cost awareness among physicians. Previous studies were mainly conducted in North America and Europe in the 1990s and 2000s and evaluated drug cost awareness among general practitioners, emergency physicians, or anesthetists [17–23]. Inadequate knowledge of costs by physicians was consistently found in these surveys [17–23]. Despite the growing need of medical responsibilization in cost control, cost awareness has not apparently improved over time. As demonstrated by our results, cost accuracy remains largely insufficient, even if we choose a quite large margin of error (±50 %) to define “correct” estimations. We also found that estimations by senior grade physicians were more accurate than their junior colleagues. This result was in contrast with previous reports that showed that the level of experience had no influence on cost awareness [20–22]. This specific influencing factor has probably been identified in our study owing to the high number of responders. Impact of professional experience on cost-consciousness is encouraging, suggesting that physicians may gradually incorporate economic considerations into their medical practices. We also found that cost estimations by female intensivists were less accurate than those of men. To the best of our knowledge, physician gender influence has never previously been documented. Our results also show that physicians have a tendency to overestimate cheap prescriptions and to underestimate expensive ones, a point consistently reported in the literature for decades [19, 21, 23]. In our study, the most expensive subgroup of prescriptions (“more than 1000 €” drugs) was the most underestimated, accounting for nearly two-thirds of the underestimation of the global amount. The presence in our questionnaire of five prescriptions exceeding 1000 € might partially explain the worrying estimates we observed. Be that as it may, high-cost drugs are now accounting for a large part of ICU budgets, and focus, more than ignorance, is required to deal with this growing major concern.

One limitation of our study might be that the inaccuracy of the estimations has not been weighted for the frequency of prescriptions or global health care management; both of these parameters are necessary to analyze the economic impact. However, through the two typical clinical cases we chose, our results allow one to indirectly appreciate the value of estimated costs in the real world. The quite small differences we observed in cost estimation per patient must be read in conjunction with the number of admissions of these patients in ICUs. For example, on the basis of a recent epidemiological study of septic shock in France [33], finding more than 50 patients yearly admitted for septic shock in each center, our results would be relevant with an approximately 10,000 € annual underestimation per ICU. Concerning hemorrhagic shock, such a dramatic amount would be reached with only four underestimations of this clinical situation. Some other limitations must be acknowledged. First, inter-hospital variability of true costs, particularly for drugs and imaging modalities, remains a reality in France, as elsewhere, that might have influenced physicians’ estimations. However, these variations can be considered as negligible when compared to the major response errors observed, especially for high-cost prescriptions. Second, we may speculate that responders were probably physicians who were the most concerned by the cost-containment issue. Nevertheless, this potential bias was substantially limited by the high response rate. Third, we have no data on how survey responders were informed on costs in each ICU; yet this factor may have affected cost estimates. Finally, the cost awareness of French physicians might have been influenced by the activity-based financing system; any transposition of our results to another health system remains uncertain.

Improving physicians’ cost awareness remains a challenge. Two important approaches can be considered: provide better information and reinforce training. Doctors appear to be predisposed to practice cost-effective medicine, but complain about problems obtaining information about costs [13]. Interventions are needed to provide reliable, easily accessible, and up-to-date cost information in everyday practice. In view of the risks of biased or inaccurate information, physicians appear to prefer academic sources or direct communication with hospital administration [21]. Another information vector could be heath information technology, increasingly used in ICUs. Associated with evidence-based decision support, computerized prescribing software providing fee data has demonstrated an efficacy to achieve cost savings [34–36]. In our study, none of the participating ICUs had adopted such a promising tool, illustrating a dramatic underutilization of cost report software. In addition, it appears essential to reinforce medical education about costs and health care management. In our study, less than 2 % of physicians had an econometrics qualification. It would be desirable for medical educators to offer more courses (in medical school, during residency, and in continuing medical education) dedicated to global health care management and general cost education [37, 38]. Professional cost-consciousness projects, which give a framework for teaching and practicing cost awareness in ICU, could also be an interesting approach [39]. As our results highlight, educational programs dedicated to cost awareness should be particularly targeted at young physicians, who are responsible for a high number of avoidable prescriptions [40, 41]. Further research should also focus on the long-term impact of cost-awareness educational programs and easier access to cost information resources.

In conclusion, this study demonstrates the alarmingly poor awareness of intensivists to costs, especially with regards to high-cost prescriptions. Considerable focus and efforts are still required to strengthen physicians’ responsibilities and to incorporate cost control in daily ICU practice.