0
Clinical Investigations in Critical Care |

A Cross-Cultural Comparison of Critical Care Delivery*: Japan and the United States FREE TO VIEW

Carl A. Sirio, MD, FCCP; Kimitaka Tajimi, MD; Nobuyuki Taenaka, MD; Yoshihito Ujike, MD; Kazufumi Okamoto, MD; Hirotada Katsuya, MD, FCCP
Author and Funding Information

*From the Department of Anesthesiology and Critical Care Medicine (Drs. Sirio), University of Pittsburgh School of Medicine, Pittsburgh, PA; Department of Emergency and Critical Care Medicine (Dr. Tajimi), Akita University School of Medicine, Akita, Japan; ICU (Dr. Taenaka), Osaka University Hospital, Osaka, Japan; Department of Emergency Medicine (Dr. Ujike), Okayama University Medical School, Okayama, Japan; Department of Emergency and Intensive Care Medicine (Dr. Okamoto), Shinsyu University School of Medicine, Matsumoto, Japan; and Department of Anesthesiology (Dr. Katsuya), Nagoya City University, Nagoya, Japan.

Correspondence to: Carl A. Sirio, MD, FCCP, University of Pittsburgh Medical Center, 612C Scaife Hall, 200 Lothrop St, Pittsburgh, PA 15213; e-mail: sirioca@anes.upmc.edu



Chest. 2002;121(2):539-548. doi:10.1378/chest.121.2.539
Text Size: A A A
Published online

Objective: To compare the utilization and outcomes of critical care services in a cohort of hospitals in the United States and Japan.

Design: Prospective data collection on 5,107 patients and detailed organizational characteristics from each of the participating Japanese study hospitals between 1993 and 1995, with comparisons made to prospectively collected data on the 17,440 patients included in the US APACHE (acute physiology and chronic health evaluation) III database.

Setting: Twenty-two Japanese and 40 US hospitals.

Patients: Consecutive, unselected patients from medical, surgical, and mixed medical/surgical ICUs.

Measurements: Severity of illness, predicted risk of in-hospital death, and ICU and hospital length of stay (LOS) were assessed using APACHE III. Japanese ICU directors completed a detailed survey describing their units.

Main results: US and Japanese ICUs have a similar array of modalities available for care. Only 1.0% (range, 0.56 to 2.7%) of beds in Japanese hospitals were designated as ICUs. The organization of the Japanese and US ICUs varied by hospital, but Japanese ICUs were more likely to be organized to care for heterogeneous diagnostic populations. Sample case-mix differences reflect different disease prevalence. ICU utilization for women is significantly lower (35.5% vs 44.8% of patients) and there were relatively fewer patients ≥ 85 years old in the Japanese ICU cohort (1.2% vs 4.6%), despite a higher per capita rate of individuals≥ 85 years old in Japan. The utilization of ICUs for patients at low risk of death significantly less in Japan (10.2%) than in the United States (12.9%). The APACHE III score stratified patient risk. Overall mortality was similar in both national samples after accounting for differences in hospital LOS, utilizing a model that was highly discriminating (receiver operating characteristic, 0.87) when applied to the Japanese sample. The application of a US-based mortality model to a Japanese sample overpredicted mortality across all but the highest (> 90%) deciles of risk. Significant variation in expected performance was noted between hospitals. Risk-adjusted ICU LOS was not significantly longer in Japan; however, total hospital stay was nearly twice that found in the US hospitals, reflecting differences in hospital utilization philosophies.

Conclusions: Similar high-technology critical care is available in both countries. Variations in ICU utilization reflect differences in case-mix and bed availability. Japanese ICU utilization by gender reflects differences in disease prevalence, whereas differences in utilization by age may reflect differences in cultural norms regarding the limits of care. Such differences provide context from which to assess the delivery of care across international borders. Miscalibration of predictive models applied to international data samples highlight the impact that differences in resource use and local practice cultures have on outcomes. Models may require modification in order to account for these differences. Nevertheless, with large databases, it is possible to assess critical care delivery systems between countries accounting for differences in case-mix, severity of illness, and cultural normative standards facilitating the design and management such systems.

Figures in this Article

International comparisons of health-care delivery focusing on clinical outcomes can assist in determining and assessing the appropriateness of health-policy goals and the adequacy with which health systems are achieving these goals. The utility of external comparisons provides insights that may not otherwise be apparent when assessing performance from within a national boundary.13 For example, there is significant variation in the use of critical care resources between industrialized nations as the number of ICU beds per capita, the percentage of beds utilized, and the number of patients receiving active ICU therapy differ.45

Despite the apparent value in understanding cross-cultural differences in health-system performance, many international comparisons of health care focus on contrasting national financing and expenditure data rather than clinical performance.6Uncoupled from clinical performance data, these statistics do not provide indepth insight into the adequacy or nature of health-care delivery. In addition, few cross-cultural comparisons of health-care delivery account for the important and complex differences that exist within the social and economic environment in which medicine is practiced across countries. Such differences can profoundly affect the efficacy and effectiveness of a health-care system.7 For example, the manner in which the health-care needs of rapidly aging populations are addressed, or the subtle biases that may exist regarding the utilization of services for selected groups of patients, can be explored using international comparative outcome data.

In a previous report,6 we presented the results of an initial comparison on intensive care in Japan and the United States in a small sample of hospitals. Our findings suggested that despite a relative cultural bias against surgical and invasive procedures in Japan when compared to the United States, and a Japanese government-mandated fee schedule that favored ambulatory services, this sample of hospitals delivered technologically advanced intensive care to their patients. When compared to the United States, ICU bed resources were limited in number and the number of low-risk or low-severity patients treated was less when compared to the United States. Differences in case-mix, disease prevalence, as well as differences in hospital admission practices and outcome by gender were observed. In addition, Japanese hospitals had the capacity to provide services, typically reserved for the ICU in the United States, on general floors but were limited in an ability to monitor differences in outcome between the floor and the ICU.

In this study, we compare utilization and outcome in 22 large Japanese tertiary-care teaching centers with 40 US hospitals providing similar types of medical and surgical high-technology care. The purpose of this study was to comprehensively understand the similarities and differences in critical care delivery by prospectively comparing the case-mix, utilization of, and outcomes from critical care services between the United States and Japan. These data can provide an alternative context from which to assess the impact of providing health services within the respective countries. To accomplish these goals, we applied a severity of illness measure developed using a US patient cohort to the Japanese data. In addition, detailed information on the organization and management practices of ICUs was assessed.

Hospital Data

We obtained information on adult ICU admissions to a stratified random and volunteer group of 42 ICUs from 40 US hospitals collected as part of the original APACHE III research effort. This hospital cohort and patient data set has been extensively described elsewhere.810

To compare patient selection, intensive-care practice, and post-ICU hospital outcome, we obtained similar data from 22 ICUs within a sample of 22 hospitals that volunteered to participate in Japan as part of an effort by the Japanese Society of Intensive Care Medicine to more fully understand the utilization of ICU beds between from 1993 to 1995. The ICU director of each study hospital responded to a detailed series of questions via a survey instrument describing the setting in which the study took place. Information collected included the policies of the ICU, teaching commitments, and organization and function of the nursing and medical staff.

Patient Data

Prospective data collected for all patients included demographic information, age, acute physiology data for 17 specific variables (temperature, heart rate, respiratory rate, BP, hematocrit, WBC count, serum sodium, albumen, bilirubin, glucose, BUN, creatinine, urine output, Pao2, Paco2, arterial pH, modified Glasgow Coma Scale) and selected chronic health information. These data were used to compute an APACHE III score. Additional hospital and patient information to calculate predictions for hospital mortality and length of stay (LOS) in this multidiagnostic data set were collected and included: ICU admission diagnosis, ICU admission source and time at source, and operative status (emergent or elective). Patient data were collected on all ICU admissions with the following exceptions: individuals < 16 years of age, burn patients, and patients with an ICU LOS of < 4 h.

Patient-specific ICU and hospital outcomes were recorded. These outcomes included ICU and hospital discharge time and date, destination following discharge (eg, home, skilled facility), ICU and hospital discharge status (alive/dead), and ICU and hospital LOS. All data were collected by nurses trained in the data collection protocol in the United States and by practicing critical care research physicians in Japan. Data were entered on a software program with internal data logic checks (APACHE Medical Systems; McLean, VA), and analyses were performed using statistical software (Version 6.0; SAS Institute; Cary, NC). Institutional Review Board approval was obtained prior to collecting data on US patients. Similar approval is not required in Japan.

Statistical Analysis

Univariate statistical comparisons of the US and Japanese patients were performed using the χ2 statistic for discrete variables. Student’s t test was used to compare differences in means of continuous measures (eg, age, APACHE III scores). Significant differences were considered at p < 0.05.

The US patient population provided the initial basis for comparing hospital mortality and LOS outcomes.9 For the Japanese patient cohort, a predicted risk of in-hospital death was estimated using a multiple logistic regression equation that was modified from the available US day 1 APACHE III risk of hospital mortality model. The model included the APACHE III acute physiology score, ICU admission source, and ICU admission diagnosis. Physiology scores were represented in the model as a continuous variable. Readmissions to the ICU were excluded from the mortality analysis to avoid counting multiple outcomes for the same patient.

The US equation includes a variable designed to control for institutional discharge triage characteristics and was not believed to be appropriate for the application to the Japanese cohort. The variable is determined for each ICU based on a statistical analysis of the length of hospital stay for all hospital survivors compared with the average stay for all ICUs based on disease and APACHE III score.

Discrimination of the final mortality logistic regression model applied to the Japanese data, as assessed by a receiver operating characteristic curve area was 0.87. To calculate a risk-adjusted mortality rate for each ICU, individual patient predictions were summed, and then a standardized mortality ratio (SMR) was calculated as the observed mortality/predicted mortality.

Logistic regression equations were generated to estimate ICU and hospital LOS using US data as a reference point for the Japanese cohort.9 These models were utilized with an expectation that the utilization of hospital resources is different in Japan when compared to the United States and that there would be wide disparities in the predicted to observed LOS outcomes when compared to the United States.

The LOS equations were based on the same patient data as described above. Once again, the excess mean adjusted length of hospital stay calculated for US patients were excluded from an estimate of the LOS of Japanese patients. In addition, institutional characteristics utilized to model US LOS, including controls for hospital size, teaching status, and geographic location, were excluded. Patients discharged to another ICU for which there was incomplete data on total length of ICU stay were excluded.

Hospital and ICU Characteristics

The US hospitals included 18 major teaching hospitals defined as university hospitals or hospitals with medical-school affiliations and at least five Accreditation Council on Graduate Medical Education-accredited residency programs. Fourteen were members of the Council of Teaching Hospitals of the American Association of Medical Colleges. The mean number of ICU beds in each hospital was 22 (range, 6 to 76 beds), and the average number of beds in study units was 13. The ICUs included 4 medical units, 8 surgical units, and 30 mixed medical/surgical units. Twenty-five hospitals had at least one step-down or intermediate-care facility.

All of the Japanese hospitals were closely linked with a medical school providing varying general and ICU-specific educational programs for medical students and graduate physicians. Sixteen of the Japanese hospitals were public facilities funded fully by the national government (n = 9) or city or prefecture governments (n = 7), and 6 hospitals were classified as private, nonprofit institutions. They were representative of institutions from diverse geographic regions. The Japanese hospitals had mean hospital and ICU bed sizes of 785 (range, 224 to 1,300 beds) and 8 (range, 6 to 16 beds), respectively. These ICUs were all mixed-function medical/surgical units. Eight hospitals had separate coronary-care units, and three others had step-down or intermediate-care units ranging in size from 8 to 20 beds.

In aggregate, the 40 US hospitals designated 4.6% of their total hospital beds for adult critical care. This compares to published estimates of ICU beds composing 6.3% of all beds in the United States.11 The apportionment of ICU beds within individual institutions in the United States sample ranged from 6 to 36%. In contrast, only 1.0% of the adult beds in the Japanese sample were designated for adult critical care, with a range of 0.56 to 2.7%. This compares with our previously reported range of 2.0 to 2.7%.7,12 Importantly, ICU directors reported all Japanese hospitals provided levels of care considered active ICU services (eg, ventilatory support, vasoactive drug therapy) in the United States on general medical and surgical hospital floors, utilizing nurses appropriately trained for such interventions, typically without the benefit of increased nurse:patient ratios associated with step-down or intermediate-care units in the United States.

With one exception, all Japanese units had full-time medical directors who were responsible for unit administration and bed allocation and triage. Similar to the United States, ICU directors had diverse backgrounds in anesthesiology, internal medicine, surgery, and emergency medicine.

In each Japanese institution, triage decisions were always reported to be made balancing the needs of those patients already in the ICU with those being assessed for admission. No formal severity-of-illness measuring system was employed when making these decisions. However, patient triage was an important part of clinical decision making, as all ICU directors estimated an ICU-bed undercapacity ranging from 10 to 50% at their respective hospitals based on ICU-bed demand.

US study ICUs displayed marked heterogeneity in the degree to which dedicated intensivists controlled both ICU triage and individual patient-care decision making. In the United States, the significant majority of ICUs had relatively open admission policies, with patient management responsibilities remaining with the admitting physician, often with the assistance of consulting physicians. In those units with an intensivist, which was not a normative standard, patient-care decisions were typically made in concert between the ICU physician and primary admitting physician.

In Japan, the unit director, in conjunction with an associated staff of dedicated intensivists, coordinated and provided the majority of care in 17 units that functioned essentially as closed ICUs. In four of the remaining units, the ICU staff was responsible primarily for respiratory and ventilatory management issues. In only one ICU did the hospital admitting physician retain primary control of the care of patients once admitted to the ICU.

The per capita availability of nurses has historically been lower in Japan when compared to the United States.13 Nevertheless, nurse:patient ratios in Japanese ICUs are similar to those found in the United States. Typically, patients receive 1:1 or 1:2 nursing care. However, nurses were responsible for a greater proportion of overall bedside work, as there are no respiratory therapists in Japan. Ancillary support personnel dedicated to caring for ventilators, hemodialysis machines, and other invasive technologies (eg, extracorporeal membrane oxygenators) are limited.

Despite the presence of an ethics committee at each of the Japanese hospitals, the ethics committees played little if any measurable role in the care and decision making of ICU patients. Further, the use of orders limiting or withdrawing care for patients with prognoses judged to be poor were limited in all centers. The use of formal protocols and procedures was also limited. Quality assurance and improvement activities were reported as informal and varied across institutions, in large measure depending on the commitment of individual clinicians to these activities. This is distinguished from ongoing quality improvement activities as a part of formal hospital and medical staff functions in the United States.

Patient Characteristics

The Japanese study population numbered 5,107 compared to 17,440 US patients. The characteristics of these two groups is summarized in Table 1 . There was nearly twice the number of men (64.5%) admitted to the Japanese ICUs when compared with the ICU admission rate for women (35.5%). The mean ± SEM age of study patients was 59.4 ± 4.3 years in the United States and 58.3 ± 2.4 years in Japan. The proportion of US patients > 85 years old was significantly larger in the United States (4.5%) than Japan (1.2%) despite population ratios that do not indicate a similar degree of disparity between the United States and Japan (United States, 12.4/1,000 people≥ 85 years old; Japan, 9.1/1,000 people ≥ 85 years old).

Postoperative patients accounted for 59.2% of the Japanese patients. This finding compares with a 42.3% prevalence of postoperative ICU admissions in the United States. Thirteen percent of patients were admitted from general medical and surgical floors in Japan, compared to 16.4% in the US sample.

The higher proportion of surgical admissions affected the distribution of diagnostic categories reflected by the most frequent reasons for ICU admission in Japan when compared to the United States (Table 2 ). This is due, in part, to known differences in the prevalence of certain disease categories, particularly GI malignancy in the Japanese population when compared to the United States.11 The high prevalence of GI malignancy also affected the relative distribution of organ-system dysfunction and failures most often responsible for ICU admission between the two countries. In the United States, the most frequent organ systems responsible for ICU admission were cardiovascular (31.7%), GI (17.9%), and neurologic (17.1%), whereas cardiovascular (34.1%), GI (22.5%), and respiratory (15.7%) systems were the most frequent in Japan.

The mean APACHE III score for all US patients was 50, with an observed in-hospital death rate of 16.5% (range, 6.4 to 40%). The mean Japanese APACHE III score was 52.5, with an observed in-hospital death rate of 18.1% (range, 39.2 to 71.7; p = not significant). The APACHE III score stratified increasing patient risk across a full spectrum of increasing severity of illness in the Japanese patient sample (Fig 1 ).

The frequency of patients classified on ICU day 1 as having a < 1% risk of death differed significantly (p < 0.001) between the two countries. The US group had a higher proportion of these ICU admissions accounting for 12.9% of the entire database, whereas in Japan 10.2% of patients were classified as having this low risk of death.

The APACHE III score stratified increasing risk over the range of severity of illness within the Japanese patient sample. The 18.1% mean predicted risk of in-hospital death and was nearly identical to the observed death rate (18.2%; p = 0.926; 95% confidence interval[ CI], 17.2 to 19.1%) with no significant gender differences. In the United States, the overall death rate was 16.5% (range, 6.4 to 40.0%). Discrimination of the US regression model applied to the Japanese cohort resulted in an receiver operating characteristic curve area of 0.87. The overall relationship between actual and predicted mortality in the Japanese ICUs is demonstrated in Figure 2 as a calibration curve. Similar to the United States, there was substantial variability in the predicted risk of death (range, 5.7 to 42.6%). In the United States, there was significant variation in the SMR (observed/predicted mortality) for the 42 units (SMR range, 0.67 to 1.25). This variation in performance was also noted in the Japanese ICUs (SMR range, 0.78 to 2.0).

The standardized ICU LOS (observed/predicted LOS) ratio in the United States varied from 0.88 to 1.21. For the Japanese ICU cohort, the overall ICU LOS ratio was 1.13 (actual ICU LOS, 6.2 days; predicted ICU LOS, 5.5 days, CI, 5.31 to 5.66 days; p = 0.000) with greater variation between ICUs (ICU LOS ratio, 0.77 to 1.66 days) when compared to the United States. Hospital LOS was significantly longer in Japan, with a mean observed LOS of 34.1 days (predicted, 17.4 days; CI, 16.9 to 19.9 days; p < 0.001).

Health-Care Delivery Systems and ICU Care

Any comparative understanding of critical care delivery must account for the social, cultural, demographic, and economic differences between the United States and Japanese health-delivery systems. The Japanese constitution guarantees “the right to maintain the minimum standards of wholesome and cultured living.” This constitutional right has led to a national health-care system, whereby three fourths of all medical care is financed through a social health insurance system. Since 1961, health-care coverage has been universal to all citizens, with cost-sharing requirements that are in aggregate less than those in the United States and a guarantee against catastrophic loss in the case of severe or prolonged illness.1416

The Japanese system labels any health-care facility with > 20 beds as a hospital. The system allows small private clinics providing varying degrees of acute-care capability to qualify under this umbrella as a hospital facility. Consequently, there are > 10,000 institutions classified as hospitals in Japan. Approximately 7,500 are privately administered and 2,500 are public facilities, nearly twice that found in the United States, which has a population double that of Japan.

Typically, it is the larger public institutions, with affiliations to the 80 Japanese medical schools, serving as major referral centers, that are responsible for a significant proportion of the critical care delivery provided throughout the country. This accounts, in part, for the relatively higher proportion of patients transferred to Japanese ICUs in this study from outside facilities (12.4%) when compared to the United States (7.2%), and the relatively higher degree of regionalized ICU care found in Japan. Contributing further to this tendency toward concentrating ICU care in larger facilities is a 1993 national government policy requiring a formalized certification of the need for highly specialized care be undertaken prior to and as part of the transfer triage process. The primary intent of this policy is to ensure that large tertiary, university-based teaching facilities remain focused on specialized care and the advancement of biomedical science. The demonstrable contribution of this policy initiative on ICU care is yet unmeasured, but can serve as a model for programs designed to coordinate and regionalize ICU services in the United States, an issue of growing importance as large health systems continue to coalesce.

Despite a system that is more likely to institutionalize the concentration of ICU resources, the geographically dispersed Japanese sample consisted of general medical/surgical ICUs. This is in contrast to the United States, where there was a greater likelihood of having ICUs specialized for the care of either medical or surgical patients. Shortell et al,10 reported a trend toward improved outcomes in those ICUs focusing on a more limited spectrum of illnesses. In addition, data1718 suggest that improved risk-adjusted outcomes for several common medical diagnoses at teaching hospitals and decreased LOS for ICU patients managed by dedicated critical care teams should focus the debate regarding the wisdom of concentrating and specializing ICU care. These data can sharpen policy debates regarding the optimal spectrum of conditions cared for within an ICU.

Outcome Modeling

In our earlier study7 of comparative ICU outcomes between Japan and the United States, we applied strictly a predictive model developed using US patients, unaltered, in a direct comparison of performance between a set of Japanese and US hospitals. This application of severity of illness and risk-stratification tools is one of several approaches to assessing performance across international boundaries, allowing a relative comparison of outcomes by setting the standard for one country as the benchmark for another. This of course does not imply that the reference benchmark is the preferred standard but rather allows for a common point for comparison.

The APACHE III score stratified increasing Japanese patient risk across the full range of increasing severity of illness. Our exclusion of variables incorporated into the US-derived APACHE III mortality and LOS predictive equations, designed as surrogate measures to capture differences in practice style and hospital utilization in the United States, highlights an alternative analytic strategy to modify models in order to account for the cultural context in which clinical care is delivered. Despite the modification of the mortality model to account partially for differences in practice style, the applied model overpredicted risk of death in all but the highest decile of risk. Possible explanations for this miscalibration included variations in patient selection that were not effectively controlled for in the risk-prediction equation, the temporal dysymetry in the data-collection periods for the analyzed data bases, and differences in the timing of hospital discharge. However, underestimation of outcome was most apparent in the prediction of hospital LOS for Japanese patients. It is standard and common practice in Japan to utilize the general floors for prolonged periods of recovery, producing hospital LOS times that are quite dissimilar to those in the United States. As Japanese hospital LOS was twice that predicted, the longer observed LOS could be expected to introduce a systematic bias producing a higher observed hospital mortality than predicted.

These differences highlight the advantages of developing country-specific predictive models designed to capture important differences in patient selection and resource utilization. Nevertheless, this does not preclude the utility of applying models developed in one country to gain insight into the relative performance of institutions within another nation as variation between predicted risk and observed outcomes are captured. This is substantiated by the broad range in SMR performance, analogous to the United States, observed in the Japanese sample suggesting large relative differences in performance across hospitals. Such comparison may improve global critical care outcomes by raising important questions regarding why differences in outcomes exist when variation between countries is uncovered.

ICU Utilization Strategies

An optimal analysis of the performance and adequacy of ICU services would account the total pool of patients for whom ICU care may be warranted (the “at risk” population), as well as those actually admitted to a critical care unit.1 Defining and measuring outcomes for this at-risk population compared to patients treated in the ICU remains problematic and poorly evaluated. The size of the at-risk group is in part dependent on the number of ICU-bed resources available. Monitoring the outcomes for the at-risk group requires a clear ability to prospectively classify such patients, a capability which in large measure does not exist. These issues have implications for this study, as the Japanese delivery system appears to have integrated an ICU philosophy to the management of critically ill patients on non-ICU floors. Lacking direct information about these patients, insight into this issue can be gained given the large difference in per capita ICU bed availability between the United States and Japan.

Thirteen percent of the ICU admissions in Japan were from general hospital floors compared to 16.4% in the United States, a 20.7% difference. Several possible reasons for this disparity may exist, including a triage strategy employed by Japanese physicians that does not easily allow patients to be admitted from the floor once a initial decision to forgo the ICU has been made, whether or not such services might otherwise be presumed to be of benefit. However, given an ability to deliver high-intensity, technologically complex services on the floor in Japanese hospitals, these data do not suggest that there is a clear adverse impact on outcomes as might be suggested by high ICU admission rates from the floor. These data suggest that the Japanese hospitals in this sample function adequately with fewer ICU resources than currently available in many US hospitals despite the subjective judgment of Japanese ICU physicians that more ICU resources are needed within their respective institutions. To definitively define the adequacy of bed resources requires an evaluation of outcome for the entire pool of patients who may potentially benefit from ICU. Such an evaluation would have significant implications for optimizing the number of ICU-bed resources in both countries.

Triage and Patient Selection for ICU Care

Within and across hospitals and health systems, important differences in triage practice, based on resource availability, personal heuristics, and local practice cultures, may lead to important differences in ICU admission policies.1920 What is considered an acceptable or appropriate ICU admission in one country may be viewed as inappropriate or extravagant in another.

Our earlier sample7 suggested that men are admitted with much greater frequency to Japanese ICUs than women. In addition, the risk-adjusted mortality for women was significantly different than that for men in Japan.4 These data confirm that a significantly higher proportion of men are admitted to Japanese ICUs when compared to women in the United States. However, this larger sample does not support the finding that risk-adjusted outcomes for men are better than those for women. The difference in ICU utilization by gender is in part a consequence of the preference to use limited ICU-bed resources for elective postoperative care and the higher incidence of GI malignancies in Japanese men compared to women for which operative resections result in postoperative ICU recovery.12 Any cultural biases regarding ICU admission by gender remains unmeasured in both the United States and Japan. Our inability to confirm earlier findings regarding differences in risk-adjusted outcome associated with gender may be a result of differences in hospital selection or sample size. A large proportion of Japanese patients admitted to the ICU are admitted from the operating room. Many of these follow elective surgery. Given the extremely small percentage of beds available, there appears to be a clear preference for ICU admission of elective surgical patients when compared to other patient groups.

In both the United States and Japan, there is increasing scrutiny of the effectiveness and outcomes related to the provision of ICU care. The need to carefully assess the impact of ICU delivery is growing in a period in which the health-care systems of both nations are attempting to maximize the utility derived from expended resources and are facing ever-aging populations. The impact of an aging population is most pronounced in Japan. How the delivery of expensive, high-technology care is structured for the elderly will have profound implications for the future costs of care in both countries.

In the United States, there is also a growing realization that we must better define suitable and sustainable goals for medical care, especially for decisions around the end of life.2124 These data provide insight into the biases that are determinative of ICU resource allocation in two different but aging societies. In Japan, where the use of “do-not-resuscitate” orders is limited as a consequence of cultural proscriptions, decisions to forgo aggressive therapy are made more often before ICU admission, thereby precluding the use of critical care resources as evidenced by the extremely small proportion of patients admitted who are > 85 years old in Japan (1.2%) when compared to the United States (4.5%). Chelluri et al25 suggest that a patient’s likely survival following ICU admission is not determined primarily by age, but rather by the degree to which the patient suffers from acute physiologic abnormality. Taken together, these data suggest that the issue of patient selection for ICU admission based on age should be scrutinized, as both health systems contemplate ways to limit spending and that objective predictions of likely outcomes be a part of the triage process.

Limitations

This study has several limitations. First, given the difference in capability to deliver ICU-like care on the general floors between the US and Japanese hospitals, we are unable to ascertain the differences in outcomes for patients treated on the floor who may have benefited from ICU care, if it had been provided. Second, the sample of hospitals in the United States was designed to be statistically representative of the majority of hospitals providing ICU care. Third, there is a time difference between when the two data sets were collected. The models are not insensitive to time. Fourth, daily severity data were not collected for the Japanese patients, precluding evaluation of the impact of low-risk patient hospital discharges and patient deaths over time. We were unable to select a random stratified sample of hospitals in Japan given the voluntary nature of this professional society-sponsored data-collection effort.

Hospitals that discharge patients later will likely report more in-hospital deaths because of the longer time during which patients could die in the hospital.26 This observation makes the Japanese mortality outcomes striking in that we may have overestimated mortality risk to a greater degree than measured. Consequently, the actual performance of the aggregate Japanese cohort could be somewhat better than we estimate. Detailed information regarding the processes of care is unavailable.

In large institutions providing ICU services, similar high-technology critical care is available in the United States and Japan. We observed variations in utilization between the United States and Japan representing differences in case-mix and bed availability. Differences in Japanese ICU utilization by gender reflect, in part, differences in the underlying prevalence of disease. Differences in utilization by age may reflect differences in cultural norms and expectations. Such distinctions provide a context from which to assess the delivery of care across international borders.

The APACHE III score stratified patients in the Japanese cohort across the full spectrum of severity of illness. In order to be well calibrated to a national sample, mortality and LOS prediction models require modification in order to account for differences in triage and practice styles between countries, highlighting the impact that differences in resource use and local practice cultures have on outcomes. By understanding the relationship between health-care spending, utilization of available resources, and outcome, one can begin to develop a clear understanding of the complex differences that exist with the cultural, social, demographic, economic, and medical environments in which efficacy and effectiveness must be evaluated. Ongoing evaluation linking health-care spending to assessment of clinical performance and outcome are essential if industrialized nations are to maintain public trust while providing adequate health care services to patients in need of expensive, life-saving therapies both inside and outside the ICU. With large databases, it is possible to assess critical care delivery systems between countries accounting for differences in case-mix, severity of illness, and cultural normative standards facilitating the design and management of critical care delivery system.

The authors and Ad Hoc Committee on Evaluation of Severity in ICU Patients of the Japanese Society of Intensive Care Medicine would like to thank the following individuals and their representative institutions for making data available for this study: Hirouki Hirasawa, MD, Chiba University School of Medicine, Chiba, Japan; Yoshimi Osakabe, MD, Fujigaoka Hospital of Showa University, Kanagawa, Japan; Choichiro Tase, MD, Fukushima Medical College, Fukushima, Japan; Akihiko Sera, MD, Hiroshima University Hospital, Hiroshima, Japan; Tatsuya Kubota, MDM Jichi Medical School, Tochigi, Japan; Akitsugu Kohama, MD, PhD, Kawasaki Medical School, Okayama, Japan; Kazufumi Okamoto, MD, Kumamoto University Medical School, Kumamoto, Japan; Akinori Zaitsu, MD, Kyushu University Hospital, Kukuoka, Japan; Naoto Nagata, MD, Miyazki Medical College Hospital, Miyazaki, Japan; Hirotada Katsuya, MD; Takefumi Azami, MD, Nagoya City University, Nagoya, Japan; Keiji Kumon, MD, National Cardiovascular Center, Osaka, Japan; Yoshihiro Yagishita, MD, National Medical Center, Tokyo, Japan; Hideki Ishihara, MD, Osaka Prefectural Habikino Hospital, Osaka, Japan; Nobuyuki Taenaka, MD, Osaka University Medical School, Osaka, Japan; Yoshihito Ujike, MD, Sapporo Medical College Hospital, Hokkaido, Japan; Seiji Yazaki, MD, Surugadai Nihon University Hospital, Tokyo, Japan; Kimitaka Tajimi, MD, Teikyo University Hospital, Tokyo, Japan; Yoshikura Haraguchi, MD, PhD, Tokyo Metropolitan Police Hospital, Tokyo, Japan; Akira Tanaka, MD, Tottori University School of Medicine, Tottori, Japan; Masahiro Shinozaki, MD, Wakayzma Medical College, Wakayama, Japan; Osamu Yamaguchi, MD, Urafune Hospital - Yokohama City University Hospital, Kanagawa, Japan; Yutaka Usada, MD, Yokohama City University School of Medicine, Kanagawa, Japan.

Abbreviations: APACHE = acute physiology and chronic health evaluation; CI = confidence interval; LOS = length of stay; SMR = standardized mortality ratio

This study was supported, in part, by the Ministry of Health (Tokyo, Japan) grants 83–1992, 127-1993, and 156-1994.

Dr. Sirio has provided ad hoc consulting services to APACHE Medical Systems.

Table Graphic Jump Location
Table 1. Demographic and Clinical Characteristics of 17,440 US and 5,107 Japanese ICU Admission Patients*
* 

Data are presented as % or mean ± SEM unless otherwise indicated.

Table Graphic Jump Location
Table 2. Most Frequent Diagnoses on ICU Admission to 40 US and 22 Japanese Hospitals
Figure Jump LinkFigure 1. Distribution of patients and the association between day 1 APACHE III score and observed in-hospital mortality rate for Japanese ICUs. The predicted mortality is based on the reference US ICU population, The APACHE III score stratified increasing risk over the range of severity of illness within the Japanese patient sample. Adequacy of calibration between observed and predicted outcome is reflected by the degree to which the observed mortality data points fall above or below the 45° reference line (ie, perfect calibration).Grahic Jump Location
Figure Jump LinkFigure 2. Calibration curve for the Japanese data set. The data points represent the relationship between actual and predicted mortality in Japanese ICUs across the full spectrum of risk of mortality. Perfect calibration would exist on the 45° line. The model tended to overpredict risk, most notably at the highest ranges of risk (Hosmer-Lemeshow goodness-of-fit test χ2, 45.5; degrees of freedom, 8; p = 000).Grahic Jump Location
Angus, DC, Sirio, CA, Clermont, G, et al (1997) International comparisons of critical care outcome and resource consumption.Crit Care Clin13,389-407. [PubMed] [CrossRef]
 
Rivera-Fernandez, R, Vazquez-Mata, G, Bravo, M, et al The APACHE III prognostic system: customized mortality predictions for Spanish ICU patients.Intensive Care Med1998;24,574-581. [PubMed]
 
Pappachan, JV, Millar, B, Bennett, ED, et al Comparison of outcome from intensive care admission after adjustment for case mix by the APACHE III prognostic system.Chest1999;115,802-810. [PubMed]
 
Abizanda, R The facilities. Miranda, DR Williams, A Loirat, P eds.Management of intensive care: guidelines for better use of resources1990,55-81 Kluwer Academic Publishers. Dordrecht, The Netherlands:
 
Zimmerman, JE, Knaus, WA, Judson, JA, et al Patient selection for intensive care: a comparison of New Zealand and United States hospitals.Crit Care Med1988;16,318-326. [PubMed]
 
Babazono, A, Hillman, AL A comparison of international health outcomes and health care spending.Int J Technol Assess Health Care1994;10,376-381. [PubMed]
 
Sirio, CA, Tajimi, KA, Tase, C, et al A comparison of critical care in Japan and the United States.Crit Care Med1992;20,1207-1215. [PubMed]
 
Knaus, WA, Wagner, DP, Draper, EA, et al The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults.Chest1991;100,1619-1636. [PubMed]
 
Knaus, WA, Wagner, DP, Zimmerman, JE, et al Variations in mortality and length of stay in intensive care units.Ann Intern Med1993;118,753-761. [PubMed]
 
Shortell, SM, Zimmerman, JE, Rousseau, DM, et al The performance of intensive care units: does good management make a difference?Med Care1994;32,508-525. [PubMed]
 
Groeger, JS, Strosberg, MA, Halpern, NA, et al Descriptive analysis of critical care units in the United States.Crit Care Med1992;20,846-863. [PubMed]
 
Tajimi, K, Shimada, Y, Nishimura, S, et al Cost containment: the pacific.Japan New Horiz1994;2,404-412
 
Health and Welfare Statistics Association. Health and welfare statistics in Japan, 1990.1990,35-84 Japan Ministry of Health. Tokyo, Japan:
 
Organization for Economic Cooperation and Development.. Outline of recent Japanese policy on pensions: the background and measure for reform. 1985; Japanese Ministry of Health and Welfare. Tokyo, Japan:.
 
Kawakita, H Healthcare services in Japan.Japan Hospitals1989;8,63-80. [PubMed]
 
Iglehart, JK Japan’s medical care system: part two.N Engl J Med1988;319,1166-1172. [PubMed]
 
for the University Health System Consortium ICU Benchmarking Study Group. Sirio, CA, Coleman, MB, McGrath, B, et al The impact of intensivists and ICU occupancy on rates of low risk-monitor-only (LRM) admissions to the ICU [abstract]. Chest. 1997;;112 ,.:16S
 
Rosenthal, GE, Harper, DL, Quinn, LM, et al Severity-adjusted mortality and length of stay in teaching and nonteaching hospitals: results of a regional study.JAMA1997;278,485-490. [PubMed]
 
Angus, DC, Linde-Zwirble, WT, Sirio, CA, et al Rural-urban differences in ICU use in Florida [abstract]. Chest. 1995;;108 ,.:105S
 
Rapoport, J, Teres, D, Barnett, R, et al A comparison of intensive care unit utilization in Alberta and Western Massachusetts.Crit Care Med1995;23,1336-1346. [PubMed]
 
Callahan, D Setting limits: medical goals in an aging society.1987,13-224 Simon and Schuster. New York, NY:
 
Murphy, DJ, Cluff, LE SUPPORT: study to understand prognosis and preferences for outcomes and risks of treatments.J Clin Epidemiol1990;43(Suppl),1S-123S
 
The SUPPORT Principal Investigators.. A controlled trial to improve care for seriously ill hospitalized patients: the study to understand prognoses and preferences for outcomes and risk of treatments (SUPPORT).JAMA1995;274,1591-1598. [PubMed]
 
Fukaura, A, Tazawa, H, Nakajima, H, et al Do-not-resuscitate orders at a teaching hospital in Japan.N Engl J Med1995;333,805-808. [PubMed]
 
Chelluri, L, Pinsky, MR, Donahoe, MP, et al Long-term outcome of critically ill elderly patients requiring intensive care.JAMA1993;269,3119-3123. [PubMed]
 
Jencks, SF, Williams, DK, Kay, TL Assessing hospital-associated deaths from discharge data: the role of length of stay and comorbidities.JAMA1988;260,2240-2246. [PubMed]
 

Figures

Figure Jump LinkFigure 1. Distribution of patients and the association between day 1 APACHE III score and observed in-hospital mortality rate for Japanese ICUs. The predicted mortality is based on the reference US ICU population, The APACHE III score stratified increasing risk over the range of severity of illness within the Japanese patient sample. Adequacy of calibration between observed and predicted outcome is reflected by the degree to which the observed mortality data points fall above or below the 45° reference line (ie, perfect calibration).Grahic Jump Location
Figure Jump LinkFigure 2. Calibration curve for the Japanese data set. The data points represent the relationship between actual and predicted mortality in Japanese ICUs across the full spectrum of risk of mortality. Perfect calibration would exist on the 45° line. The model tended to overpredict risk, most notably at the highest ranges of risk (Hosmer-Lemeshow goodness-of-fit test χ2, 45.5; degrees of freedom, 8; p = 000).Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1. Demographic and Clinical Characteristics of 17,440 US and 5,107 Japanese ICU Admission Patients*
* 

Data are presented as % or mean ± SEM unless otherwise indicated.

Table Graphic Jump Location
Table 2. Most Frequent Diagnoses on ICU Admission to 40 US and 22 Japanese Hospitals

References

Angus, DC, Sirio, CA, Clermont, G, et al (1997) International comparisons of critical care outcome and resource consumption.Crit Care Clin13,389-407. [PubMed] [CrossRef]
 
Rivera-Fernandez, R, Vazquez-Mata, G, Bravo, M, et al The APACHE III prognostic system: customized mortality predictions for Spanish ICU patients.Intensive Care Med1998;24,574-581. [PubMed]
 
Pappachan, JV, Millar, B, Bennett, ED, et al Comparison of outcome from intensive care admission after adjustment for case mix by the APACHE III prognostic system.Chest1999;115,802-810. [PubMed]
 
Abizanda, R The facilities. Miranda, DR Williams, A Loirat, P eds.Management of intensive care: guidelines for better use of resources1990,55-81 Kluwer Academic Publishers. Dordrecht, The Netherlands:
 
Zimmerman, JE, Knaus, WA, Judson, JA, et al Patient selection for intensive care: a comparison of New Zealand and United States hospitals.Crit Care Med1988;16,318-326. [PubMed]
 
Babazono, A, Hillman, AL A comparison of international health outcomes and health care spending.Int J Technol Assess Health Care1994;10,376-381. [PubMed]
 
Sirio, CA, Tajimi, KA, Tase, C, et al A comparison of critical care in Japan and the United States.Crit Care Med1992;20,1207-1215. [PubMed]
 
Knaus, WA, Wagner, DP, Draper, EA, et al The APACHE III prognostic system: risk prediction of hospital mortality for critically ill hospitalized adults.Chest1991;100,1619-1636. [PubMed]
 
Knaus, WA, Wagner, DP, Zimmerman, JE, et al Variations in mortality and length of stay in intensive care units.Ann Intern Med1993;118,753-761. [PubMed]
 
Shortell, SM, Zimmerman, JE, Rousseau, DM, et al The performance of intensive care units: does good management make a difference?Med Care1994;32,508-525. [PubMed]
 
Groeger, JS, Strosberg, MA, Halpern, NA, et al Descriptive analysis of critical care units in the United States.Crit Care Med1992;20,846-863. [PubMed]
 
Tajimi, K, Shimada, Y, Nishimura, S, et al Cost containment: the pacific.Japan New Horiz1994;2,404-412
 
Health and Welfare Statistics Association. Health and welfare statistics in Japan, 1990.1990,35-84 Japan Ministry of Health. Tokyo, Japan:
 
Organization for Economic Cooperation and Development.. Outline of recent Japanese policy on pensions: the background and measure for reform. 1985; Japanese Ministry of Health and Welfare. Tokyo, Japan:.
 
Kawakita, H Healthcare services in Japan.Japan Hospitals1989;8,63-80. [PubMed]
 
Iglehart, JK Japan’s medical care system: part two.N Engl J Med1988;319,1166-1172. [PubMed]
 
for the University Health System Consortium ICU Benchmarking Study Group. Sirio, CA, Coleman, MB, McGrath, B, et al The impact of intensivists and ICU occupancy on rates of low risk-monitor-only (LRM) admissions to the ICU [abstract]. Chest. 1997;;112 ,.:16S
 
Rosenthal, GE, Harper, DL, Quinn, LM, et al Severity-adjusted mortality and length of stay in teaching and nonteaching hospitals: results of a regional study.JAMA1997;278,485-490. [PubMed]
 
Angus, DC, Linde-Zwirble, WT, Sirio, CA, et al Rural-urban differences in ICU use in Florida [abstract]. Chest. 1995;;108 ,.:105S
 
Rapoport, J, Teres, D, Barnett, R, et al A comparison of intensive care unit utilization in Alberta and Western Massachusetts.Crit Care Med1995;23,1336-1346. [PubMed]
 
Callahan, D Setting limits: medical goals in an aging society.1987,13-224 Simon and Schuster. New York, NY:
 
Murphy, DJ, Cluff, LE SUPPORT: study to understand prognosis and preferences for outcomes and risks of treatments.J Clin Epidemiol1990;43(Suppl),1S-123S
 
The SUPPORT Principal Investigators.. A controlled trial to improve care for seriously ill hospitalized patients: the study to understand prognoses and preferences for outcomes and risk of treatments (SUPPORT).JAMA1995;274,1591-1598. [PubMed]
 
Fukaura, A, Tazawa, H, Nakajima, H, et al Do-not-resuscitate orders at a teaching hospital in Japan.N Engl J Med1995;333,805-808. [PubMed]
 
Chelluri, L, Pinsky, MR, Donahoe, MP, et al Long-term outcome of critically ill elderly patients requiring intensive care.JAMA1993;269,3119-3123. [PubMed]
 
Jencks, SF, Williams, DK, Kay, TL Assessing hospital-associated deaths from discharge data: the role of length of stay and comorbidities.JAMA1988;260,2240-2246. [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
Guidelines
Critical care in pregnancy.
American College of Obstetricians and Gynecologists | 7/10/2009
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543