This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Ochroch, E. A. Articles by Schwartz, S. J. Search for Related Content PubMed PubMed Citation Articles by Ochroch, E. A. Articles by Schwartz, S. J. Related Collections Critical Care Monitoring (Non-cardiac)

---

CRITICAL CARE AND TRAUMA

The Impact of Continuous Pulse Oximetry Monitoring on Intensive Care Unit Admissions from a Postsurgical Care Floor

Anesthesia and Cardiopulmonary Services, University Health Systems East; Department of Anesthesiology, University of Pennsylvania, Philadelphia, Pennsylvania

Address correspondence to E. Andrew Ochroch, MD, Assistant Professor, Department of Anesthesiology, University of Pennsylvania Health System, 680 Dulles Building, 3400 Spruce Street, Philadelphia, PA 19104. Address e-mail to ochrocha{at}uphs.upenn.edu.

	Abstract

Top Abstract Introduction Methods Results Discussion References

 
Continuous pulse oximetry (CPOX) has the potential to increase vigilance and decrease pulmonary complications and thus decrease intensive care unit (ICU) admissions. In a randomized nonblinded study of 1219 subjects we compared the effects of CPOX and standard monitoring on the rate of transfer to an ICU from a 33-bed postcardiothoracic surgery care floor. There was no difference in the rate of ICU readmission between the CPOX and standard monitor groups. Despite older age and comorbidity, estimated cost to time of censoring (enrollment to completion of the study) was less in the monitored patients who required ICU transfer than in the unmonitored patients who required ICU transfer (mean estimated cost difference of $28,195; P = 0.04). Use of CPOX altered the reasons that patients were transferred to an ICU but did not affect the rate of transfer. The duration, and thus estimated cost, of ICU stay was significantly less in the CPOX-monitored group. The potential for CPOX to allow for early intervention, or perhaps prevention of pulmonary complications, needs to be explored. Routine CPOX monitoring did not reduce transfer to ICU, mortality, or overall estimated cost of hospitalization, and it is unclear if there is any real benefit from the application of this technology in patients on a general care floor who are recovering from cardiothoracic surgery.

	Introduction

Top Abstract Introduction Methods Results Discussion References

 
Medical errors and their impact on patient morbidity, mortality, and the estimated cost of care are receiving intense scrutiny (1–9). An area of growing concern is the potential for errors caused by increased demands on physicians and nurses, particularly from being required to care for more patients. Although many complications are idiosyncratic, their severity and associated morbidity may often be significantly reduced through prompt detection (10). Although mostly untested, advanced monitoring devices offer the potential to increase patient safety and improve patient outcomes by detecting problems early, when they are easily treated and before they cause significant morbidity, while lessening caregiver burden.

Effective monitoring technologies should, at a minimum, measure an etiological variable that can be acted upon to improve patient outcome. Ideally, they should detect an incipient problem before it causes significant morbidity or requires intensive intervention. Acting on this problem early enough should prevent the subsequent adverse outcome. Such monitoring technologies should be accurate, precise, and reproducible; they should be easy to implement without disrupting routine care processes, easy to consistently apply and interpret, and they should not increase risk to the patient. Finally, such technologies should be cost-effective or cost-saving (11). Such monitoring devices have particular potential when applied to the highest risk patients, such as postsurgical patients recently transferred from the cardiothoracic intensive care unit (CTICU) (12).

Intensive care unit (ICU) admission after major surgical procedures and traumatic injury is common (13). Appropriate availability and use of ICU resources is a priority for most large medical centers. Unplanned admissions and readmissions to ICUs from operating rooms and general care areas cause significant disruption in intrahospital patient care and are associated with high estimated costs and excess morbidity and mortality (14–16). In the case of cardiothoracic (CT) surgical services, ready access to ICU beds is crucial for a high-quality system and optimal clinical outcomes (12).

Despite reduced invasiveness of many CT procedures, a postoperative ICU stay is still the norm for many patients. Once discharged from the ICU, many patients are placed on "telemetry" floors to obtain continuous electrocardiographic (ECG) monitoring. ECG telemetry monitoring was adapted from the cardiac medical ICU, where it is thought to reduce late mortality from myocardial infarction (17). Although cardiac arrhythmias are common, particularly after cardiac surgery, there are no studies documenting the clinical effectiveness of routine ECG monitoring in the postoperative setting (15,16).

In contrast, there is considerable evidence that respiratory compromise is a more common and significant cause of postoperative admission and readmission to ICUs than cardiac dysrhythmias (15,16). Systems for monitoring arterial oxygen saturation via pulse oximetry (POX) have been used routinely in operating rooms for at least two decades. After their rapid adoption in the operating room, these monitors have become standards of practice in postanesthetic care units (PACUs) and ICUs. Recently, continuous pulse oximetry (CPOX) using centralized data displays similar to ECG telemetry stations has been introduced.

After an internal quality improvement process identifying respiratory compromise as the most common indication for ICU readmission after CT surgery at our institution, we performed a randomized controlled trial to examine whether the use of CPOX with centralized monitoring on the post-CT surgery care floor would reduce ICU admission and readmission rates. Further, we wanted to assess the impact of CPOX on the estimated cost of care for this population. We hypothesized that CPOX would reduce unplanned respiratory and total admissions to the ICU for CT postoperative patients from the general surgical care floor and decrease length of ICU readmission and that it would do so in a cost-effective manner.

	Methods

Top Abstract Introduction Methods Results Discussion References

 
This investigator-initiated, randomized clinical trial was funded by Nellcor (a division of Tyco Healthcare). After approval by the IRB at the University of Pennsylvania, a prospective, nonblinded, randomized, clinical trial was initiated to assess the impact of routine CPOX on ICU transfer, ICU and hospital length of stay, and estimated cost for postoperative CT patients. Nellcor ‘s central station network CPOX system Oxinet® II was installed on a single 33-bed hospital floor that primarily accepted patients recovering from cardiac and thoracic surgery. Standardized orientation and training of floor nurses was undertaken to ensure proper use and interpretation of CPOX. A preliminary prospective cohort study of 260 consecutive admissions indicated that use of CPOX resulted in a 30% reduction in transfers from the study floor to the ICU as compared with historic controls and a shorter duration of stay in the ICU after transfer. Based on these data, power analysis determined that a sample size of 2000 was required to be 80% certain of detecting a 33% reduction in transfers to an ICU at the 95% confidence level (P < 0.05). Secondary end-points included duration of time between admission to study floor and hospital discharge, length of ICU readmission, estimated cost of ICU and total hospital stay, and mortality for study subjects assigned to CPOX versus usual care.

Enrollment for the study began in June 1999. All patients admitted to the study floor Monday through Friday between 10 am and 5:30 pm were eligible for entry into the study unless they had previously been randomized (i.e., if they were enrolled, transferred to an ICU, and then readmitted to the study floor, they were not eligible for repeat randomization).

Patients were randomized to CPOX ("monitored") or standard ("unmonitored") floor care using sequential sealed envelopes containing randomly generated group assignments. Randomization was overseen by a single study research nurse not involved in clinical care of study subjects. At the time a patient agreed to participate, the study research nurse was contacted, the next envelope was opened, and the randomization code was assigned. Because of the nature of the intervention, the nurses on the study floor were not blinded as to group assignment. However, study investigators who conducted the outcome assessment were blinded as to study group.

In the unmonitored group, POX was intermittently assessed as per hospital protocol according to clinical need based on nursing or physician judgment. Three stand-alone, noncentralized POX monitors were available for longer periods of continuous monitoring in the unmonitored group on physician order. Thus, the study tested the clinical and estimated economic impact of routine CPOX use versus "typical" use of POX in current practice. More than 95% of all patients in both study groups received concurrent centralized ECG telemetry monitoring. A study investigator reviewed the random assignments daily to ensure that patients received the monitoring appropriate to their assigned study group.

Patients in the CPOX monitored group used a commercially available third-generation bedside CPOX unit (N-3000; Nellcor Puritan-Bennett, Pleasanton, CA) connected via a hard-wired system to a centralized data display unit (Oxinet® II). There were 3 separate nursing stations on the study floor, each with the capability of monitoring up to 11 beds simultaneously. System alarms were active at the bedside unit and at the central display. Each bedside unit was capable of storing 24 h of CPOX data. The centralized monitoring station was capable of storing 24 h of CPOX data for each oximeter. These data were collected each day using an interface developed locally and were stored using proprietary software (SCORE®, Nellcor) for later analysis. The SCORE software was designed to allow analysis of the data in terms of the degree of hemoglobin desaturation, duration of hemoglobin desaturation, fidelity of the waveform, number of desaturation episodes, and pulse rate behavior. Beyond standard in-service training on the CPOX technology, no special instruction or suggested response to the CPOX data was provided to clinical personnel. Once enrolled, patients were followed until they were transferred to another floor, transferred to an ICU, discharged home, or died. Patients who were transferred to an ICU from the study floor were followed until transfer out of that ICU. Patients were not reenrolled in the study (i.e., if they were enrolled, transferred to an ICU, and then readmitted to the study floor, they were not eligible for repeat randomization). They were followed according to the standard of care.

During the initial phase of the study, hospital policy required transfer to an ICU of any patient receiving an IV infusion of amiodarone. Twenty-two study patients (9 monitored, 13 unmonitored) were identified who transferred to the ICU during this period. A blinded chart review was conducted by one of the authors (EAO) to determine if these patients were transferred because of deterioration in their health or solely for amiodarone administration for treatment of postoperative atrial fibrillation without hemodynamic consequences. Two patients (one in each group) had no documentation of significant medical compromise, and their data were treated as if they had not returned to the ICU. Estimated cost data for both patients were censored at the time of transfer to the ICU.

Clinical information for each study patient was entered into a prospectively designed Access (Microsoft, Redmond, WA) database. Data sources included the daily clinical record, the SCORE recording of the CPOX information, and a blinded standardized chart review completed after patient discharge. ASA physical status was determined from the subject's preoperative assessment by their anesthesiologist. Hospital discharge diagnoses and ambulatory visit diagnoses were reviewed to calculate a Charlson comorbidity score (18,19).

Estimated cost data were determined using the Hospital of the University of Pennsylvania's proprietary billing system; these data included all hospital services but excluded physicians' professional fees. Patient records were searched and charges were identified and aggregated from admission to discharge from the hospital. Charges were determined from our proprietary SMS billing system (SMS Systems, Malvern, PA), which includes line-item charge detail for all procedures and includes supplies, nursing, pharmacy, radiology, and allied health care/support staff. It does not include professional fees, except when the fee is included in a specific billable procedure (i.e., ECG and chest radiograph). Charges were calculated for four time intervals: 1) study enrollment to censoring (leaving the study floor because of hospital discharge, transfer to another floor, or, if transferred to an ICU, on discharge from the ICU); 2) study enrollment to first disposition (charges accumulated during time on the study floor from enrollment to leaving the study floor because of hospital discharge, transfer to another floor or transfer to an ICU; 3) duration of ICU admission for patients transferred from the study floor to the ICU; and 4) total hospital stay from admission into the hospital to discharge home, including any and all ICU stays and operating room estimated costs but not professional fees. Cost centers such as respiratory, radiology, and ICU were used to confirm dates of transfer and discharge, as determined by chart review. Charges were converted to estimated costs using the institution's overall global cost-to-charge ratio of 0.40. Because data collection continued beyond 1 yr, estimated costs were discounted by 4% per year, where appropriate.

The primary outcomes examined were the number of subjects transferred to an ICU and estimated costs from enrollment to censoring (time of leaving the study floor because of hospital discharge, transfer to another floor, or, if transferred to an ICU, on discharge from the ICU). Secondary analysis examined estimated costs and duration of the ICU readmission, reason for transfer to the ICU, and estimated cost and duration of the entire hospitalization.

Demographic data were analyzed using one-way analysis of variance, the Mann-Whitney U-test, and Fisher's exact test (two-tailed) for continuous, ordinal, and proportional data, respectively. Differences in the rate of ICU readmission for the CPOX and control groups were assessed using Student's t-test and regression analysis (analysis of covariance) to adjust for the effect of possible confounders.

Estimated cost data were analyzed using both Student's t-tests and the Mann-Whitney U-test. The Student's t-test on untransformed estimated cost data tests the arithmetic mean and is the most appropriate approach for health policy decisions that must consider the total estimated cost of treatment (20). The right-skewed distribution of estimated cost data is typical because a few patients incur high costs resulting from complications, reoperation, or extended hospital stays. Although these data are not normally distributed, Student's t-tests are sufficiently robust given a sufficient sample size (20,21). Student's t-tests were used to assess the statistical significance of differences in mean estimated cost between the monitored and unmonitored groups from time of enrollment to censoring (primary outcome), from time of enrollment to discharge, and for differences in estimated ICU costs between the groups for those patients transferred to an ICU. The Mann-Whitney U-test tests the median of the data, and represents a typical estimated cost (20). The Mann-Whitney U-test was applied to the same analyses as the Student's t-test.

A bootstrap analysis of the estimated cost for ICU stay and the estimated cost to censoring was performed for the 93 patients who returned to the ICU (22). We produced 1000 samples of 93 with replacement and used Student's t-tests to analyze the differences in estimated costs between monitored and unmonitored groups (Satterthwaite correction for unequal variances).

	Results

Top Abstract Introduction Methods Results Discussion References

 
Over the 26 months of enrollment, 1219 subjects were enrolled of approximately 8300 patients who met eligibility criteria; 22 patients refused to participate in the study. Enrollment was less than projections primarily because only one enrollment nurse was available from 10 am to 5:30 pm Monday through Friday and there was a decrease in the number of available patient beds during the study as a result of a nursing shortage. Demographics of enrolled patients are shown in Table 1. The monitored and unmonitored groups did not differ by age, gender, race, surgical service, location before transfer to the study floor, number of days in the hospital before enrollment, or Charlson score.

As shown in Table 1, subjects transferred to an ICU who were monitored with CPOX were sicker, as evidenced by a higher Charlson score (3 versus 2; P = 0.005), and older (72 versus 64; P = 0.056). There was a trend for the monitored subjects to be transferred to the ICU earlier (day 3 versus day 4; P = 0.091). Forty-six percent of all patients had cardiac surgery, 34% had thoracic surgery, 5% had vascular surgery, 10% had general surgery, and the remaining 5% were from all other services.

Rates of readmission to the ICU (Table 1) were similar in the monitored and unmonitored groups. Of the 93 patients (8% of all subjects) who were transferred to the ICU after enrollment, 40 were in the monitored group (of 589; 6.7%) and 53 were in the unmonitored group (of 630; 8.5%) (P = 0.33). Subjects transferred to the ICU did not differ from those not transferred to the ICU ("Discharged") in terms of age, gender, race, or surgical service (Table 1). Starting in the CTICU before transfer to our study floor was significantly associated with return to an ICU from our study floor (odds ratio [OR], 2.1; 95% confidence interval [CI], 1.3 – 4.9; P = 0.001). The reasons for transfer back to the ICU (determined by blinded review of the ICU transfer notes) differed between monitored and unmonitored groups (Table 2), with more pulmonary events (versus cardiac or "other") in the unmonitored group compared with the CPOX monitored group (27 versus 8; P = 0.003).

Mean estimated costs from enrollment to censoring were lower in the monitored group compared with unmonitored patients ($15,481 versus $18,713; P = 0.038). Despite their older age and greater comorbidity, estimated cost to time of censoring was lower in the monitored group that required ICU transfer than for the unmonitored patients who required ICU transfer (mean estimated cost difference of $28,195; P = 0.04) (Table 3). Similarly, estimated costs for ICU care alone were lower in the monitored group than in the unmonitored group (mean estimated cost difference of $23,004; P = 0.04). These findings are particularly robust as similar differences were found when comparing median outcomes. Adjustment for covariates (age, race, gender, Charlson score, ASA status) did not significantly affect estimated cost outcomes.

The bootstrap analysis further confirmed the robustness of our finding of reduced estimated ICU costs in the monitored group. The mean estimated cost of ICU stay was $23,262 less for the monitored than the unmonitored group (P < 0.0001), with the monitored group incurring lower estimated ICU costs than the unmonitored group more than 97.5% of the time (95% CI always showed estimated ICU cost savings). Similarly, the monitored group's median estimated cost to censoring was $28,200 less than the unmonitored group's estimated cost (P < 0.0001), with the 95% CI always indicating savings.

The use of CPOX did not impact duration of stay in the hospital or total estimated cost of hospitalization when examining the entire cohort. Routine CPOX and usual care groups had similar numbers of days from enrollment to discharge from study, numbers of days from enrollment to discharge from the hospital, estimated costs while on the study floor, and estimated costs for the entire hospital stay.

There were 14 in-hospital deaths in each group. We were concerned that very early or very late deaths (prolonged hospital stay) might act as outliers for length of stay and estimated costs. The effects of the deaths on the study outcomes were examined by reexamining the outcomes without these patients' data considered (sensitivity analysis). The deaths did not affect outcomes or produce/enhance differences between groups.

Because coming from the CTICU to our study floor was significantly associated with return to an ICU from our study floor (OR, 2.1; 95% CI, 1.3 – 4.9; P = 0.0012), an exploratory subgroup analysis for this cohort was undertaken. Of the 690 patients initially transferred to the study floor from the CTICU, 70 (10.1%) returned to the ICU after enrollment in the study. Among this subgroup of patients transferred from an ICU to the study floor and subsequently readmitted to an ICU, there was a mean estimated cost difference between the monitored and unmonitored groups of $21,328 (P = 0.033) but there was no difference in estimated cost to time of censoring or in total hospital estimated cost. This is supported by the difference in median estimated costs for ICU care of $18,668 (P = 0.045).

	Discussion

Top Abstract Introduction Methods Results Discussion References

 
We had hypothesized that CPOX would reduce admission or readmission to an ICU by allowing earlier intervention in postoperative respiratory complications, allowing these to be avoided or treated on the floor. This randomized clinical trial of third-generation CPOX technology did reduce respiratory readmissions to an ICU post-CT surgery, but it is unclear why there was a concomitant increase in cardiac complications requiring transfer to an ICU in the monitored group (Table 2). It is possible that CPOX monitoring increases overall nursing vigilance, resulting in increased nonrespiratory ICU transfers. If appropriate, such transfers may contribute to the benefits of CPOX observed in this study. If such transfers represent inappropriately aggressive care, then there is the potential for routine CPOX monitoring to further reduce ICU readmissions with further training of nurses. This issue thus requires further study.

If there is a real benefit of CPOX in reducing ICU transfer, the lack of reduction in the absolute rate of return to an ICU by CPOX use can also be considered a dilutional effect. The large group of patients who did well regardless of the monitoring overwhelms any beneficial effect of monitoring on the much smaller group who may have benefited from the monitoring. This is similar to the perioperative data, which show a decreased rate of hypoxemia when POX is used (23) but no change in rare outcomes (myocardial infarction, stroke, and death) from the use of POX (24,25). We attempted to determine if such a dilutional effect was a factor by examining a more clinically "at-risk" population within our cohort: subjects who came to the study floor from the CTICU, who would presumably be more likely to benefit from CPOX monitoring. As discussed above, fewer CPOX patients in the "prior location CTICU" group returned to the ICU (31 versus 39; P = 0.075). However, given the limited statistical power of this study, this approximate 30% reduction did not reach conventional standards of statistical significance. Together with the fact that this was a post hoc analysis, this must be considered a hypotheses to be further examined in future research.

CPOX use was associated with decreased mean estimated costs for monitored versus unmonitored patients ($15,481 versus $18,713; P = 0.038) from enrollment to censoring. Monitored patients who required readmission to the ICU appeared to be recognized earlier and subsequently required a shorter period of time in the ICU, thus incurring less estimated ICU costs. As noted above, CPOX reduced estimated costs/charges of ICU care and estimated costs/charges for the censored time period but not total hospital estimated costs/charges. This may reflect the fact that the intervention was often a short-lived component of a much longer and complex hospitalization, the variance of which within and between groups was quite large, this either compromising our ability to demonstrate overall savings or, alternatively, swamping any such savings. This is typical of single-point interventions: the entire clinical care pathway needs to change to produce a shortened, and thus less expensive, hospital stay (26,27).

Study enrollment was less than the number estimated by power analysis as necessary to determine our primary end-point of a 33% decrease in the rate of transfer to an ICU. Underpowering a study often results in trends of uncertain significance. Further, our enrollment times from 10 am to 5:30 pm may have biased our patient selection into a nonrepresentative pool. It is possible that patients entering the floor between those times represent healthier scheduled transfers from the ICU or patients doing well in the postanesthesia care unit and transferred directly to the floor. We would not have been able to capture patients who were "bumped out" of an ICU by a sicker patient. This overall effect would have been to make the group overall healthier and less likely to benefit from more intensive monitoring.

This unblinded trial did not attempt to alter nursing or physician behavior, which may have caused underestimation or over-estimation of the impact of the monitoring because treatment regimens based on CPOX results were not prescribed. Monitoring may lead to increased costs from greater nursing time as the result of false alarms or action taken on clinically unimportant periods of hypoxemia. Our data do not suggest increased estimated costs associated with increased detection of such hypoxemic episodes (i.e., increased respiratory treatments or chest radiographs). It is possible that time spent with patients who may experience such episodes may distract caregivers from more appropriate patient care. Further, as noted above, hypervigilance resulting from CPOX use may have appropriately, or inappropriately, resulted in increased transfers to the ICU for cardiac reasons.

Although we found no clear benefit in reduction of ICU transfers, we did find indications that CPOX may have changed patient care based on fewer pulmonary causes for return to an ICU in the monitored group. However, it is unclear why this group would have an increase in cardiac complications and no overall decline in the rate of ICU admission. Logically, it would seem that the patient group at greatest risk for postoperative pulmonary complications–-those having had CT surgery and a previous ICU stay (CTICU before transfer to the study floor)–-could potentially benefit most and are thus a reasonable target for further investigation. Furthermore, CPOX-monitored subjects had less estimated costs from time of enrollment into the study until censoring and less estimated costs of ICU readmissions.

Although this application of CPOX technology did not decrease the rate of transfer to an ICU, several potential benefits were demonstrated that warrant rigorous future study. Routine CPOX monitoring altered the reasons for ICU readmission (Table 2), reducing the number of patients transferred to the ICU for pulmonary reasons, and thus may indicate real benefit for the surgical patients who are at great risk for postoperative pulmonary complications. We suspect that this benefit was a result of earlier detection and treatment of pulmonary events that otherwise would have led to ICU transfer, but this hypothesis needs to be tested. It would have been, and will be, interesting to explore aggressive pulmonary interventions based on rates and patterns of desaturation as detected by CPOX. This may help to address the expense and frequent mortality in CT patients who are transferred back to an ICU (14–16).

In conclusion, in this population of patients, use of CPOX was associated with reduced postoperative ICU admission for pulmonary complications, with reduced estimated costs of care from study enrollment to censoring and with reduced estimated costs during ICU stay after transfer from the study floor. Routine CPOX monitoring was not associated with overall decreased transfer to ICU, mortality, or total estimated costs of hospitalization.

Although the results of this study are not conclusive, the data suggest that there may be a benefit for a population at high risk for postoperative pulmonary complications. Future research will be needed to examine and expand on these findings. If these findings are confirmed, such routine monitoring postoperatively may become a standard of care in patients at high risk of postoperative pulmonary complications.

	Footnotes

 
Supported, in part, by an unrestricted grant from Nellcor/Tyco Healthcare, Inc., and NIH grant K23 HD40914-02.

Accepted for publication October 6, 2005.

Reprints will not be available.

	References

Top Abstract Introduction Methods Results Discussion References

 

1. Leape LL. Reporting of adverse events. N Engl J Med 2002;347:1633–8.[Free Full Text]
2. Kohn LT CJ, Donaldson MS (eds). To err is human: building a safer health system. Washington, DC: National Academy Press, 2000.
3. Hughes CM. How many deaths are due to medical errors? JAMA 2000;284:2187.[Free Full Text]
4. Honig P, Phillips J, Woodcock J. How many deaths are due to medical errors? JAMA 2000;284:2188–9.
5. Brennan TA. The Institute of Medicine report on medical errors–could it do harm? N Engl J Med 2000;342:1123–5.[Free Full Text]
6. Anderson RE. How many deaths are due to medical errors? JAMA 2000;284:2189.[Free Full Text]
7. Newhall C. The Institute of Medicine report on medical errors. N Engl J Med 2000;343:664.
8. Richardson WC, Berwick DM, Bisgard JC. The Institute of Medicine report on medical errors. N Engl J Med 2000;343:665.[Free Full Text]
9. Richardson WC, Berwick DM, Bisgard JC, et al. The Institute of Med Report on Medical Errors: misunderstanding can do harm. Quality of Health Care in America Committee. Med Gen Med (computer file): Medscape General Med 2000;E42.
10. Silber JH, Williams SV, Krakauer H, Schwartz JS. Hospital and patient characteristics associated with death after surgery: a study of adverse occurrence and failure to rescue. Med Care 1992;30:615–29.[Web of Science][Medline]
11. MacGregor DA, Royster RL. Perfecting the imperfect. Crit Care Med 1999;27:2311–2.[Web of Science][Medline]
12. Cohn WE, Sellke FW, Sirois C, et al. Surgical ICU recidivism after cardiac operations. Chest 1999;116:688–92.[Abstract/Free Full Text]
13. Sirio CA, Martich GD. Who goes to the ICU postoperatively? Chest 1999;115:125S–9S.[Abstract/Free Full Text]
14. Kramer DJ. Intensive care unit frequent fliers: morbidity and estimated cost. Crit Care Med 2001;29:207–8.[Web of Science][Medline]
15. Rosenberg AL, Watts C. Patients readmitted to ICUs: a systematic review of risk factors and outcomes. Chest 2000;118:492–502.[Abstract/Free Full Text]
16. Rosenberg AL, Hofer TP, Hayward RA, et al. Who bounces back? Physiologic and other predictors of intensive care unit readmission. Crit Care Med 2001;29:511–8.[Web of Science][Medline]
17. Whalen RE, Ramo BW, Wallace AG. The value and limitations of coronary care monitoring. Prog Cardiovasc Dis 1971;13:422–36.[Medline]
18. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83.[Web of Science][Medline]
19. Deyo C, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epi 1992;45:613–9.[Web of Science][Medline]
20. Thompson SG, Barber JA. How should estimated cost data in pragmatic randomised trials be analysed? BMJ 2000;320:1197–200.[Free Full Text]
21. Coyle D. Statistical analysis in pharmacoeconomic studies: a review of current issues and standards. Pharmacoeconomics 1996;9:506–16.[Web of Science][Medline]
22. Barber JA, Thompson SG. Analysis of estimated cost data in randomized trials: an application of the non-parametric bootstrap. Stats Med 2000;19:3219–36.[Web of Science][Medline]
23. Moller JT, Jensen PF, Johannessen NW, Espersen K. Hypoxaemia is reduced by pulse oximetry monitoring in the operating theatre and in the recovery room. Br J Anaesth 1992;68:146–50.[Abstract/Free Full Text]
24. Moller JT, Johannessen NW, Espersen K, et al. Randomized evaluation of pulse oximetry in 20,802 patients: II. Perioperative events and postoperative complications. Anesthesiology 1993;78:445–53.[Web of Science][Medline]
25. Moller JT, Pedersen T, Rasmussen LS, et al. Randomized evaluation of pulse oximetry in 20,802 patients: I. Design, demography, pulse oximetry failure rate, and overall complication rate. Anesthesiology 1993;78:436–44.[Web of Science][Medline]
26. New clinical paths often need process redesign to finish the estimated cost-saving/CQI job. Health Care Estimated Cost Reengineering Report 1997;2:9–13.
27. Shorr AF. An update on cost-effectiveness analysis in critical care. Curr Opin Crit Care 2002;8:337–43.[Medline]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via Web of Science (3) Citing Articles via Google Scholar Google Scholar Articles by Ochroch, E. A. Articles by Schwartz, S. J. Search for Related Content PubMed PubMed Citation Articles by Ochroch, E. A. Articles by Schwartz, S. J. Related Collections Critical Care Monitoring (Non-cardiac)