Secondary Logo

Journal Logo

A Novel Index of Hypoxemia for Assessment of Risk During Procedural Sedation

Niklewski, Paul J., PhD*†; Phero, James C., DMD; Martin, James F., PhD*; Lisco, Steven J., MD

doi: 10.1213/ANE.0000000000000371
Technology, Computing, and Simulation: Research Report
Free
SDC

BACKGROUND: Procedural sedation is essential for many procedures. Sedation has an excellent safety profile; however, it is not without risks. Assessment of risk using clinical outcomes in clinical studies is difficult due to their rare occurrence. Therefore, surrogate end points are frequently used in a clinical study in lieu of clinical outcomes. As a clinician integrates multiple aspects of a physiological variable to determine potential risk, a surrogate end point should consider a similar approach. In this study, we identified and tested the appropriateness of a new surrogate end point that may be used in clinical studies, area under the curve of oxygen desaturation (AUCDesat). A review of patient sedation records by anesthesiologists was conducted to assess its relationship to the anesthesia professional perception of risk.

METHODS: This study was a post hoc analysis and assessment of perceived risk by anesthesiologists. It consisted of 13 U.S.-trained board-certified anesthesiologists ranking physiological variables as indicators of risk and then reviewing 204 records from 3 completed sedation studies involving the SEDASYS® System. After review, each anesthesiologist assigned a Likert score based on his or her perception of risk for oversedation-related sequelae in each record. These scores were analyzed to determine their relationship to desaturation presence/absence, duration, depth, number of events, and AUCDesat that incorporates each component.

RESULTS: Anesthesiologists ranked arterial oxygenation to be the most important factor in assessing risk post hoc (mean rank of 4.69 of 5, P = 0.0007 compared with next highest ranked factor—respiratory rate, N = 13). AUCDesat was better correlated to the Likert scores (rs = 0.85) when compared with the individual elements of AUCDesat, binary assessment of desaturation (rs = 0.73), desaturation depth (rs = −0.70), desaturation duration (rs = 0.70), and incidence of desaturations (rs = 0.55) (all 4 comparisons versus rs = 0.85, P < 0.0001).

CONCLUSIONS: Anesthesiologists determined arterial oxygenation to be the most important physiological variable in assessing sedation risk and the potential for adverse clinical outcomes. AUCDesat, a composite index that incorporates duration, incidence, and depth of oxygen desaturation, was better correlated to the Likert scores. AUCDesat, given that it is a single numerical variable, is an ideal end point for assessment of risk of adverse clinical outcomes in clinical sedation studies. Future studies using AUCDesat and actual physiological outcomes may be useful in further defining this end point.

From *Ethicon Endo-Surgery, Inc., Cincinnati, Ohio; Department of Neuroscience, University of Cincinnati, Cincinnati, Ohio; Department of Anesthesiology, University of Cincinnati Medical Center, Cincinnati, Ohio.

Steven J. Lisco, MD, is currently affiliated with the Department of Anesthesiology, University of Nebraska Medical Center, Omaha, Nebraska.

Accepted for publication June 5, 2014.

Funding: This research was funded by Ethicon Endo-Surgery, Inc., Cincinnati, Ohio, USA.

Conflicts of Interest: See Disclosures at the end of the article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Address correspondence and reprint requests to Paul Niklewski, PhD, Ethicon Endo-Surgery, Inc., 4545 Creek Rd., Cincinnati, OH 45209. Address e-mail to pniklews@its.jnj.com.

Procedural sedation helps manage a patient’s pain, fear, and anxiety, allowing for improved comfort and compliance for painful or uncomfortable diagnostic or therapeutic procedures. Moderate IV sedation is not without risk1; however, rates of adverse outcomes are rare. For example, mortality due to anesthesia was assessed at a rate of approximately 1/1,000,000 per year.2 For office-based procedures such as oral surgery, the mortality rate has been estimated at 1/1,733,055.3 This implies that a simple superiority study could require a sample size exceeding one million subjects. Clearly, such a study would be prohibitive to run in a prospective manner.

To understand the risk of different sedation treatments, it is advantageous to use a surrogate end point when assessing safety of procedural sedation in a clinical sedation study. Classically, patient injury is a direct result from respiratory depression leading to cardiac compromise.4 Surrogate end points commonly used for assessing injury are oxygen desaturation, apnea, hypotension, and bradycardia. Loss of consciousness in the continuum of sedation5 has also been used at times as a surrogate safety end point. However, loss of consciousness as a surrogate end point is problematic in its application because it is not in the direct causal pathway of injury. To properly assess adverse outcomes using surrogate end points, the end points must be in the causal pathway of injury.6

This study assessed existing surrogate end points as well as an alternate surrogate end point, area under the curve of oxygen desaturation (AUCDesat). Specifically, the study assessed which end points an anesthesiologist is most likely to consider when reviewing physiology from a procedural sedation record. This included evaluating which end points may be most appropriate for use in assessing risk during sedation and how to apply the surrogate end point in a clinical study.

Back to Top | Article Outline

METHODS

The study consisted of 13 U.S.-trained, board-certified anesthesiologists with a minimum of 10 years in practice, who reviewed sedation records consisting of patient physiology from completed IRB-approved sedation clinical studies executed during the development and assessment of the SEDASYS® System.a The informed consent for those studies had provisions allowing for retrospective analysis of all data collected. Anesthesiologists were asked to review anonymous patient data and assign their perceived risk for each patient to each record on a Likert scale. The record order and results from each anesthesiologist were blinded using a random identifier generated with R (R Core Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; ISBN 3-900051-07-0, URL.

Back to Top | Article Outline

Sedation Record Assessment

Each anesthesiologist received a set of 204 sedation records from completed clinical studies (a record example is shown in Fig. 1). Each anesthesiologist was provided basic information about the patient population for the sedation records. Patients were ASA physical status I, II, or III with a body mass index <35. Patients received minimal/moderate sedation. All patients received supplemental oxygen. Patient physiological variables were collected using a sidestream capnometer, pulse oximeter, and noninvasive arterial blood pressure monitor, respectively. Patient data were deidentified and compiled from 3 completed clinical sedation studies involving the SEDASYS System. One hundred twelve sedation records contained 5 physiological variables (saturation, respiration rate, end-tidal carbon dioxide (EtCO2), heart rate, and arterial blood pressure), and 92 records contained only 1 of the physiological variables. The single variable records were included in this study to allow assessment of the anesthesiologists’ scoring associated with just that specific variable unbiased by other contextual data. First, the anesthesiologists ranked their perceived importance (1 = least important; 5 = most important) of each physiological variable. Next, each anesthesiologist ranked his or her perception of the patient’s risk for oversedation-related sequelae for each sedation record using an 11-point Likert scale (0 = no risk of sequelae; 10 = highest risk of sequelae) for each of the 204 records.

Figure 1

Figure 1

Back to Top | Article Outline

Appropriateness of End Points as Surrogates

As discussed, surrogate end points need to be in the causal pathway of injury.6 Demonstrating a correlation alone is not sufficient when considering the appropriateness of a surrogate end point. In the case of oversedation, the primary injury is an insult to the central nervous system or the cardiovascular system. In a study of the American Society of Anesthesiologists Closed Claims Database, >40% of claims associated with monitored anesthesia care (MAC) involved mortality or central nervous system injury.4 From the same analysis, the primary mechanisms of injury were (in descending order of occurrence) respiratory events, cardiovascular events, equipment failures, events related to regional blockade, inadequate anesthesia or patient movement, and medication-related events.

Looking at respiratory depression, the primary mechanism of injury from MAC,4 the patient may present with apnea, hypercarbia, impaired gas exchange, or a combination of the 3.7 In a nonintubated patient, this is best measured by observing the patient’s respiratory rate and EtCO2.8 If a respiratory depression event continues, the patient will manifest both hypoxemia and a cardiovascular response. This patient response can be assessed using a patient’s peripheral oxygen saturation, arterial blood pressure, and heart rate.7 Given that the primary mechanisms of injury from MAC are respiratory depression and/or cardiovascular events, assessment for desaturation, apnea, hypotension, hypercarbia, and bradycardia is reasonable.

While broad application of these end points could be applied, it is necessary to assess the impact of the procedure or the drugs on the end points. For example, hypotension may not be an ideal surrogate for certain procedures or medications. During a colonoscopy, there is a potential for a vasovagal reaction that can result in hypotension and bradycardia,9 confounding the use of hypotension as a surrogate for oversedation. Drugs may also complicate the use of hypotension as a surrogate. Ketamine, a common anesthetic, has a known complication of increased arterial blood pressure.10 The effect of ketamine on arterial blood pressure confounds the application of hypotension as a surrogate for oversedation. Clearly, the procedure and the drugs being assessed need to be considered when selecting an appropriate surrogate.

Back to Top | Article Outline

End Point Definitions

The study end points defined a priori and analyzed were oxygen desaturation, apnea, hypercarbia, hypotension, and bradycardia. The definitions of the end points were taken from prior sedation studies.11,12 Oxygen desaturation was defined as an episode of arterial oxygen saturation <90% for a duration of ≥15 seconds; apnea was defined as no respiratory rate activity for at least 30 seconds; hypercarbia as an EtCO2 ≥50 mm Hg for a duration of ≥30 seconds; hypotension as the lower value of a systolic measure <80 mm Hg or 80% of baseline; and bradycardia as the lower value of a heart rate <50 bpm or 80% of baseline for a duration of ≥30 seconds.

Back to Top | Article Outline

Area Under the Curve of Oxygen Desaturation

We evaluated the validity of a unique clinical risk index called AUCDesat (Fig. 2). AUCDesat has been used in a prior clinical study of the SEDASYS System.11 This measure was calculated for each record by combining the incidence, duration, and magnitude of a patient’s oxygen desaturation. AUCDesat is an integrated numerical value representing the total area under 90% arterial oxygen saturation. For example, if a patient’s oxygen desaturation was 85% for a period of 30 seconds, the AUCDesat for that duration is 150 (seconds %). This is calculated by multiplying the saturation difference of 5% (90%–85%) by the 30-second duration.

Figure 2

Figure 2

Back to Top | Article Outline

Statistical Analysis

All statistical significance tests and tests of correlation were performed in Systat version 13 from Systat Software, Inc. (San Jose, CA), except as specified below. The physiological parameter-ranking mean presented is an arithmetic mean. The 95% confidence interval for the parameter rankings was calculated using Systat.

Correlations were calculated using a Spearman rank-sum bootstrap for each factor.b This was done using SYSTAT 13’s bootstrap resampling method for correlation calculations. The number of samples for the bootstrap was 1000, and the size of each sample was the number of records available with the parameter in question (minimum sample size was 102). The Mersenne twister random number generator was used. The reported confidence intervals for the correlations are 95% using the percentile method.

A random effects analysis was performed to determine the effect of the anesthesiologist reviewer and the record on the Likert score. This analysis was done using SYSTAT 13. The software code is provided in the Online Only Supplement (Supplemental Digital Content, http://links.lww.com/AA/A930). The fixed effect was the incidence of desaturation, desaturation events, depth of desaturation, duration of desaturation, or AUCDesat. The random effects were the anesthesiologist and the record.

R was used to determine the statistical significance between the correlation scores from AUCDesat to the other desaturation elements. Specifically, the paired.r(xy,xz,yz,n) (psych toolbox) function was used. The desaturation data were assumed to be dependent; therefore, xy represented the correlation of AUCDesat to the mean Likert score, xz represented the correlation of the specific desaturation element to the mean Likert score, and yz represented the correlation between AUCDesat and the specific desaturation element. For example, comparing AUCDesat’s correlation to the Likert scores (rs = 0.85) to the incidence of desaturation’s correlation to the Likert scores (rs = 0.70) would be calculated as paired.r(0.85,0.73,0.884,153). The rs between AUCDesat and the incidence of desaturation was 0.884. One hundred fifty-three was the number of evaluated records with saturation data. The resultant t value is 5.79, which is P < 0.0001 using 152 degrees of freedom.

Back to Top | Article Outline

RESULTS

Results of the Physiological Parameter Ranking and Record Assessment

The anesthesiologist’s rankings of the 5 physiologic variables are shown in Table 1. Arterial oxygen saturation was ranked as most important with a mean ranking of 4.69 (95% confidence interval: 4.40–4.98, P = 0.0007 compared with respiratory rate, N = 13). Respiratory rate’s mean ranking of 3.15 (95% confidence interval: 2.34–3.97, N = 13) did not differ significantly (P = 0.20) when compared with arterial blood pressure, the variable with the third highest estimated rank. Because SpO2 was the only measure to be statistically different from the others, the remainder of the analysis focused on surrogate end points that used the patient’s oxygen saturation as its source.

Table 1

Table 1

A random effects analysis was performed to understand the role of the anesthesiologist and the specific records on the calculated correlation and ranked values. The random effects analysis provided the same calculated P values for these measures. For each desaturation element (including AUCDesat), the type III fixed effect remained statistically significant. Anesthesiologists, evaluated as a random effect, were predominantly statistically significant. This was due to each anesthesiologist having a different perception of the absolute Likert score value. Two anesthesiologists did not trend well with AUCDesat. The first-ranked apnea as the most important variable and the Likert scores trended more with respiration compromise than desaturation. The second anesthesiologist was extremely sensitive to any desaturation, treating it as a binary factor, resulting in low sensitivity to AUCDesat. For each of the records, those with a desaturation had a statistical significance on the Likert score and to a lesser degree those with other adverse physiology.

Figure 3A shows the mean scores of the record reviews completed by all 13 anesthesiologists. The distribution is left skewed, reflecting a lower Likert score for the presented records. Figure 3B and C shows the variability among the anesthesiologists’ Likert scores. They illustrate the difference between the maximal and minimal scores and the standard deviation of the scores for each record, respectively. The difference between the maximal and minimal scores was sometimes as much as 9 (reflecting scores on the opposite ends of the scale). The standard deviation still shows score variability but not as much as the difference between the maximum and minimum scores.

Figure 3

Figure 3

Evaluation of the correlation between the mean Likert score from the records and the components of desaturation is listed in Table 2. The assessed components were presence/absence of desaturation, the minimal SpO2 value during the desaturation, the duration of the desaturation, the number of desaturation events in a record, and the calculated AUCDesat. The mean Likert score is associated well with oxygen desaturation as measured by the pulse oximeter and minimal value and duration. The number of events was reasonably correlated, although it was the lowest of the individual components. AUCDesat was most highly correlated.

Table 2

Table 2

Scatter plots for the different components of a desaturation (incidence, depth, and duration) as well as AUCDesat (Fig. 4) show that AUCDesat was better correlated to the mean Likert score compared with the individual desaturation elements. AUCDesat had a Spearman correlation value of rs = 0.85. When compared with the individual elements of AUCDesat, binary assessment of desaturation (rs = 0.73), desaturation depth (rs = −0.70), desaturation duration (rs = 0.70), and incidence of desaturations (rs = 0.55), it is statistically different for all 4 elements (P < 0.0001). This is the P value for the difference between the correlations, which is why rs = 0.85 is statistically significant, while the confidence intervals for each correlation value may overlap. The smoothed lines in the figure highlight the relationship between the variable and the perceived risk. The smoothing function used was a Loess smoother (local smoothing technique with tricube weighting and polynomial regression). To further validate the P value, the linear mixed model used to understand the role of random effects (reviewer and record) also provided a result of P < 0.0001 when comparing AUCDesat with each of the individual desaturation elements (pairwise).

Figure 4

Figure 4

Back to Top | Article Outline

Use of AUCDesat as a Statistical End Point

AUCDesat, by design, is a non-normal distribution when used during procedural sedation. There will be a large collection of zero values because most patients are not expected to experience a desaturation. Those who do desaturate will likely have an AUCDesat that is low, given that most desaturations are mild. As a result, there is a left-skewed distribution. This can be problematic because application of a traditional t test requires normality of the data. Plotting the histogram of the data collected in the records, the non-normal distribution is clearly evident (Fig. 5A). Using a logarithmic data transform of all nonzero values, the distribution can be transformed to a normal distribution (Fig. 5B). If only the nonzero values are evaluated, then a mixed model should be used to ensure that the influence of the zero values is included. A more sophisticated approach would be to use a zero-inflated log-normal model. Li et al.13 present an example of using a zero-inflated log normal on modeled data. One of the benefits of the approach described is that it is hierarchical, allowing for the inclusion of random variables to understand inter- and intradata variations. Alternatively, the data may be assessed using the Kruskal-Wallis in place of a 1-way analysis of variance or a Wilcoxon signed-rank test instead of a paired t test.14 The log(AUCDesat) (Fig. 6) was highly correlated (0.90) with the mean Likert scores, further supporting using log as a transform for data analysis.

Figure 5

Figure 5

Figure 6

Figure 6

Back to Top | Article Outline

DISCUSSION

AUCDesat could be an ideal surrogate end point for clinical studies comparing sedation modalities. It appears that anesthesiologists, when reviewing a patient’s physiology, are including the incidence, depth, and duration of a desaturation. By incorporating duration, depth, and incidence of oxygen desaturation into a single composite index, it is possible that this new surrogate will better reflect the anesthesiologist’s perception of patient risk during procedural sedation.

A limitation of this study is that AUCDesat was correlated to an anesthesiologist’s perception of the risk of oversedation-related sequelae rather than to actual patient outcomes. However, as a consideration, it is possible to create a hypothetical grouping for AUCDesat in an attempt to both put context around the AUCDesat and create a stratification for future studies to validate.

First, the AUCDesat values can be grouped into 3 categories: low, medium, and high. To create these groups, the log(AUCDesat) data were clustered using K-Means splitting with euclidean distance for the distance metric. The first group had 27 records, with a mean log(AUCDesat) of 1.87 (1.34–2.15). The second group resulted in a group of 45 records, with a mean log(AUCDesat) score of 2.45 (2.17–2.82). The third group contained 17 records, with a mean log(AUCDesat) of 3.31 (3.00–3.85). Taking the inverse log, the ranges in the 3 groups were (22–141), (147–664), and (923–7040). Approximating these values, we can create the 3 groups. A low Likert score (0–5) is associated with an AUCDesat of <150. A medium Likert score (6–8) is associated with an AUCDesat of 150 to 900 (seconds %), and a high Likert score (9–10) is associated with an AUCDesat of >900 (seconds %).

These 3 groups can be associated with established groupings of desaturation and potential injury. LaManna15 describes sustained hypoxia and its potential effects on the central nervous system using 4 states of desaturation. The first is nominal saturation, with PaO2 of 12 kPa (90 mm Hg) or more. Mild hypoxic exposure is defined as a PaO2 of >6 kPa (45 mm Hg) and <12 kPa (90 mm Hg). At this level, LaManna15 describes that normal physiological mechanisms appear to manage the hypoxia and there is seldom residual tissue damage. If the PaO2 is 4 kPa (30 mm Hg) to 6 kPa (45 mm Hg), LaManna15 describes this as moderate hypoxia, where the physiological mechanisms partially compensate for the lack of oxygen but there is commonly permanent damage. Lastly, he describes severe hypoxia as a PaO2 of <4 kPa (30 mm Hg), where neuronal degeneration results. These regions of hypoxia are shown on the oxyhemoglobin curve shown in Figure 7.

Figure 7

Figure 7

Converting these PaO2 values to saturation (SaO2), LaManna’s groups can be transformed into groups defined by percent oxygen saturation of hemoglobin. Assuming a nominal PaCO2 and temperature, the nominal saturation group has a saturation value ≥97%. Mild hypoxic exposure is between 80% and 97%. Moderate hypoxic exposure is between 57% and 80%. Severe hypoxic exposure occurs at saturation <57%.

These calculated saturation groups can be compared with the AUCDesat groups based on the anesthesiologists’ Likert scores. The first and most notable difference is that LaManna15 considered a 97% saturation (PaO2 = 12 kPa, 90 mm Hg) as the threshold for hypoxia. This threshold is perhaps too stringent for a clinical study, given the patient variability seen for baseline saturation and recognizing that LaManna15 was addressing sustained saturation values <97%. For the intent of this index, a saturation of 90% was shown to be an adequate threshold for hypoxia, as demonstrated by the strong correlation of AUCDesat and the anesthesiologists’ mean Likert score. Starting with the AUCDesat stratification based on an anesthesiologist’s perception of risk (low 0–150; medium 150–900; high >900), the durations of hypoxia required to align with LaManna’s classification can be determined. Figure 8 shows an approximation of the different regions and some example desaturations, providing clinical insight into the AUCDesat ranges.

Figure 8

Figure 8

Back to Top | Article Outline

Statistical Considerations

In the application of AUCDesat, the lack of normality in the data needs to be considered. As mentioned above, it is likely that there will be a disproportionate amount of AUCDesat scores equaling 0 from a study. Therefore, a statistical analysis plan should consider the method of managing the non-normal distribution of AUCDesat. For small sample size studies, a planned logarithmic transformation of the AUCDesat may be appropriate. For larger sample sizes (N > 30 per group, depending on the final distribution of the data), the central limit theorem applies, and an analysis of variance may be used in the analysis of the data. For this reason, special consideration needs to be made regarding the sample size of the population and the influence of covariates.

Back to Top | Article Outline

CONCLUSIONS

It appears that anesthesiologists, when reviewing a patient’s physiological record for perceived risk, are integrating the number of oxygen desaturations, the depth of the desaturations, and the duration of the desaturations. Therefore, with respect to an end point for sedation clinical studies, it seems intuitive to include all the constituent components of a desaturation into a singular numerical index. AUCDesat, by incorporating incidence, depth, and duration of oxygen desaturation, is a new surrogate end point that will better reflect the anesthesiologist’s perception of risk, when compared with the more traditional measures of incidence alone. Additional studies will further define the role of oxygen desaturation and how AUCDesat correlates to actual physiological outcomes.

Back to Top | Article Outline

DISCLOSURES

Name: Paul J. Niklewski, PhD.

Contribution: This author contributed to the original study design, original data analysis, data interpretation, drafting of the manuscript, review, revisions, and final approval of the manuscript.

Attestation: Paul J. Niklewski attests to having approved the final manuscript and to the integrity of the original data and the analysis reported in this manuscript. Paul J. Niklewski is the archival author.

Conflicts of Interest: Paul J. Niklewski declares his conflict of interest as an employee of Ethicon Endo-Surgery, Inc., a Johnson & Johnson Company, sponsor of this research. Ethicon Endo-Surgery, Inc., was the developer of the SEDASYS System, and this manuscript presents AUCDesat, the primary end point from the pivotal study for the SEDASYS System.

Name: James C. Phero, DMD.

Contribution: This author contributed to survey participation, interpretation of data, critical review and revisions of the manuscript, and final approval of the manuscript.

Attestation: James C. Phero attests to having approved the final manuscript.

Conflicts of Interest: James C. Phero consulted for EES but is not currently consulting. This author did not receive any fees for his participation in the research or as an author on this manuscript.

Name: James F. Martin, PhD.

Contribution: This author contributed to the original study design, original data analysis, data interpretation, drafting of the manuscript, review, revisions, and final approval of the manuscript.

Attestation: James F. Martin attests to having approved the final manuscript and to the integrity of the original data and the analysis reported in this manuscript.

Conflicts of Interest: James F. Martin declares his conflict of interest as an employee of Ethicon Endo-Surgery, Inc., a Johnson & Johnson Company, sponsor of this research. Ethicon Endo-Surgery, Inc., was the developer of the SEDASYS System, and this manuscript presents AUCDesat, the primary end point from the pivotal study for the SEDASYS System.

Name: Steven J. Lisco, MD.

Contribution: This author contributed to survey participation, interpretation of data, critical review and revisions of the manuscript, and final approval of the manuscript.

Attestation: Steven J. Lisco attests to having approved the final manuscript.

Conflicts of Interest: Steven J. Lisco consulted for EES, but is not currently consulting. This author did not receive any fees for his participation in the research or as an author on this manuscript.

This manuscript was handled by: Maxime Cannesson, MD, PhD.

Back to Top | Article Outline

ACKNOWLEDGMENTS

Julie Kesler, BS, MBA, provided assistance with study execution, data collection and analysis, preparation of this article, and editorial assistance. Betsy Weise provided assistance with data collection.

Back to Top | Article Outline

FOOTNOTES

a This study was a retrospective review of patient clinical data from IRB- approved studies for the SEDASYS System. The Informed Consent for those studies had provisions allowing for retrospective analysis of all data collected. Because this research involved patient data, we complied with the HIPAA Privacy Rule (46 CFR Part 160; Part 164 [subparts a,e]) during the execution of this retrospective analysis. For this exploratory analysis, a minimum of 10 anesthesiologists was targeted with a minimum of 200 record reviews. As there is no prior data with which to size the study, the sample size was not statistically calculated. Records were sent to 13 anesthesiologists, with the expectation that all may not respond. The minimum target for records reviewed by each anesthesiologist was 200; 2% were added (204) to ensure that the minimum was met.
Cited Here...

b Thirteen anesthesiologists provided 204 values each, resulting in a theoretical total of 2652 Likert scores (ranging from 0 to 10). There were a total of 6 missing records, resulting in an actual number of 2646 records assessed. The maximum number of missed records by an anesthesiologist was 3, with 3 others missing 1 record each.
Cited Here...

Back to Top | Article Outline

REFERENCES

1. Barlow D, Bell G, Chalmers A, Charlton J Implementing and Ensuring Safe Sedation Practice for Healthcare Procedures in Adults. 2001 London, UK: U.K. Academy of Medical Royal Colleges and Their Faculties:1–21
2. Li G, Warner M, Lang BH, Huang L, Sun LS. Epidemiology of anesthesia-related mortality in the United States, 1999-2005. Anesthesiology. 2009;110:759–65
3. D’Eramo EM, Bontempi WJ, Howard JB. Anesthesia morbidity and mortality experience among Massachusetts oral and maxillofacial surgeons. J Oral Maxillofac Surg. 2008;66:2421–33
4. Bhananker SM, Posner KL, Cheney FW, Caplan RA, Lee LA, Domino KB. Injury and liability associated with monitored anesthesia care: a closed claims analysis. Anesthesiology. 2006;104:228–34
5. ASA. Continuum of Depth of Sedation: Definition of General Anesthesia and Levels of Sedation/Analgesia. 2009 Schaimburg, IL: American Society of Anesthesiologists;
6. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann Intern Med. 1996;125:605–13
7. Hemmings HC, Hopkins PM Foundations of Anesthesia: Basic Sciences for Clinical Practice. 20052 edn Philadelphia, PA: Mosby Elsevier;
8. Krauss B, Hess DR. Capnography for procedural sedation and analgesia in the emergency department. Ann Emerg Med. 2007;50:172–81
9. Herman LL, Kurtz RC, McKee KJ, Sun M, Thaler HT, Winawer SJ. Risk factors associated with vasovagal reactions during colonoscopy. Gastrointest Endosc. 1993;39:388–91
10. Truven Health Analytics. . Ketamine (Injection route). 2013 Ann Arbor, MI: Truven Health Analytics;
11. Pambianco DJ, Vargo JJ, Pruitt RE, Hardi R, Martin JF. Computer-assisted personalized sedation for upper endoscopy and colonoscopy: a comparative, multicenter randomized study. Gastrointest Endosc. 2011;73:765–72
12. Pambianco DJ, Whitten CJ, Moerman A, Struys MM, Martin JF. An assessment of computer-assisted personalized sedation: a sedation delivery system to administer propofol for gastrointestinal endoscopy. Gastrointest Endosc. 2008;68:542–7
13. Li N, Elashoff DA, Robbins WA, Xun L. A hierarchical zero-inflated log-normal model for skewed responses. Stat Methods Med Res. 2011;20:175–89
14. McDonald JH Handbook of Biological Statistics. 20092nd ed. Baltimore, MD Sparky House Publishing
15. LaManna JC. Hypoxia in the central nervous system. Essays Biochem. 2007;43:139–51
16. Kelman GR. Digital computer subroutine for the conversion of oxygen tension into saturation. J Appl Physiol. 1966;21:1375–6
17. Severinghaus JW. Simple, accurate equations for human blood O2 dissociation computations. J Appl Physiol Respir Environ Exerc Physiol. 1979;46:599–602

Supplemental Digital Content

Back to Top | Article Outline
© 2014 International Anesthesia Research Society