Penalized Spline-Involved Tree-based (PenSIT) Learning for estimating an optimal dynamic treatment regime using observational data

Dynamic treatment regimes are a set of time-adaptive decision rules that can be used to personalize treatment across multiple stages of care. Grounded in causal inference methods, dynamic treatment regimes identify variables that differentiate the treatment effect and may be used to tailor treatments across individuals based on the patient's own characteristics - thereby representing an important step toward personalized medicine. In this manuscript we introduce Penalized Spline-Involved Tree-based Learning, which seeks to improve upon existing tree-based approaches to estimating an optimal dynamic treatment regime. Instead of using weights determined from the estimated propensity scores, which may result in unstable estimates when weights are highly variable, we predict missing counterfactual outcomes using regression models that incorporate a penalized spline of the propensity score and other covariates predictive of the outcome. We further develop a novel purity measure applied within a decision tree framework to produce a flexible yet interpretable method for estimating an optimal multi-stage multi-treatment dynamic treatment regime. In simulation experiments we demonstrate good performance of Penalized Spline-Involved Tree-based Learning relative to competing methods and, in particular, we show that Penalized Spline-Involved Tree-based Learning may be advantageous when the sample size is small and/or when the level of confounding of the outcome is high. We apply Penalized Spline-Involved Tree-based Learning to the retrospectively-collected Medical Information Mart for Intensive Care dataset to identify variables that may be used to tailor early fluid resuscitation strategies in septic patients.

A retrospective cohort study of predictors and interventions that influence cooperation with mask induction in children

BACKGROUND

Uncooperative pediatric mask induction is linked to perioperative anxiety. Although some risk factors for uncooperative inductions have been reported, there are no large cohort studies that identify intrinsic patient characteristics associated with cooperation.

AIM

The primary aim was to identify patient characteristics associated with cooperative mask inductions. The secondary aim was to determine whether preoperative interventions were associated with increased cooperation.

METHODS

This retrospective cohort study included patients 2-11 years old and ASA class I-IV who underwent mask induction. Our primary outcome of interest was cooperation with mask induction, which was correlated against the Induction Compliance Checklist. The variables analyzed for association with cooperation were age, sex, ASA class, class of surgery, preferred language, and race. Interventions examined for association with induction cooperation included premedication with midazolam, exposure to distraction technology, parental presence, and the presence of a Child Life Specialist. Multivariate mixed-effects logistic regression was used to assess the relationship between patient characteristics and cooperation. A separate multivariate mixed-effects logistic regression was used to examine the association between preoperative interventions and cooperation.

RESULTS

9692 patients underwent 23 474 procedures during the study period. 3372 patients undergoing 5980 procedures met inclusion criteria. The only patient characteristic associated with increased cooperation was age (OR 1.20, p-value 0.03). Involvement of Child Life Specialists was associated with increased cooperation (OR 4.44, p-value = 0.048) while parental/guardian presence was associated with decreased cooperation (OR 0.38, p-value = 0.002).

CONCLUSION

In this cohort, increasing age was the only patient characteristic found to be associated with increased cooperation with mask induction. Preoperative intervention by a Child Life Specialists was the sole intervention associated with improved cooperation.

A multifaceted quality improvement project improves intraoperative redosing of surgical antimicrobial prophylaxis during pediatric surgery

BACKGROUND

Accurate intraoperative antibiotic redosing contributes to prevention of surgical site infections in pediatric patients. Ensuring compliance with evolving national guidelines of weight-based, intraoperative redosing of antibiotics is challenging to pediatric anesthesiologists.

AIMS

Our primary aim was to increase compliance of antibiotic redoses at the appropriate time and appropriate weight-based dose to 70%. Secondary aims included a subset analysis of time compliance and dose compliance individually, and compliance based on order entry method of the first dose (verbal or electronic).

METHODS

At a freestanding, academic pediatric hospital, we reviewed surgical cases between May 1, 2014, and October 31, 2017 requiring antibiotic redoses. After an institutional change in cefazolin dosing in May 2015, phased interventions to improve compliance included electronic countermeasures to display previous and next dose timing, an alert 5 minutes prior to next dose, and weight-based dose recommendation (September 2015). Physical countermeasures include badge cards, posting of guidelines, and updates to housestaff manual (September 2015). Statistical process control charts were used to assess overall antibiotic redose compliance, time compliance, and dose compliance. The chi-square test was used to analyze group differences.

RESULTS

A total of 3015 antibiotic redoses were administered during 2341 operative cases between May 1, 2014, and October 31, 2017. Mean monthly compliance with redosing was 4.3% (May 2014-April 2015) and 73% (November 2015-October 2017) (P < 0.001). Dose-only compliance increased from 76% to 89% (P < 0.001), and time-only compliance increased from 4.9% to 82% (P < 0.001). After implementation of countermeasures, electronic order entry compared with verbal order was associated with higher dose compliance, 90% vs 86% (P = 0.015).

CONCLUSION

This quality improvement project, utilizing electronic and physical interventions, was effective in improving overall prophylactic antibiotic redosing compliance in accordance with institutional redosing guidelines.

Pediatric Postoperative Pulse Oximetry Monitoring During Transport to the Postanesthesia Care Unit Reduces Frequency of Hypoxemia

BACKGROUND

The standard use of pulse oximetry during the transport of postoperative patients from the operating room (OR) to the postanesthesia care unit (PACU) is not routinely practiced. A study was conducted to determine if the frequency of hypoxemia on admission to the PACU decreased after implementation of routine use of transport pulse oximeters for postoperative patients being transferred to the PACU.

METHODS

In this prospective cohort study, which was conducted at an academic pediatric hospital, the primary outcome measure was the frequency of hypoxemic events on arrival to the PACU.

RESULTS

A total of 506 patients in the preintervention phase and 597 in the postintervention phase met the inclusion criteria. Six hypoxemic events on arrival to the PACU were identified in preintervention phase versus zero in the postintervention period, p = 0.009. Use of oxygen monitors during transport from the OR to the PACU increased from 0% to 100%, p < 0.0001, in the postintervention phase. The median duration of unmonitored time during transport decreased from 272 seconds to 13 seconds, p < 0.0001. Of the 605 patients who met the inclusion criteria for sustainment audits-conducted 18 months after the postimplementation evaluation-99.8% were transported to the PACU with a pulse oximeter, and there were zero reported hypoxemic patients on PACU admission.

CONCLUSION

The routine use of portable oxygen monitoring when transferring patients from the OR to the PACU is a low-cost, noninvasive safety measure that should be considered at any institution performing pediatric general anesthesia.

A Strategic Approach to Implementation of Medical Mentorship Programs

BACKGROUND

Mentors influence medical trainees' experiences through career enhancement and psychosocial support, yet some trainees never receive benefits from involved mentors.

OBJECTIVE

Our goals were to examine the effectiveness of 2 interventions aimed at increasing the number of mentors in training programs, and to assess group differences in mentor effectiveness, the relationship between trainees' satisfaction with their programs given the presence of mentors, and the relationship between the number of trainees with mentors and postgraduate year (PGY).

METHODS

In group 1, a physician adviser funded by the graduate medical education department implemented mentorships in 6 residency programs, while group 2 involved a training program with funded physician mentoring time. The remaining 89 training programs served as controls. Chi-square tests were used to determine differences.

RESULTS

Survey responses from group 1, group 2, and controls were 47 of 84 (56%), 34 of 78 (44%), and 471 of 981 (48%, P = .38), respectively. The percentages of trainees reporting a mentor in group 1, group 2, and the control group were 89%, 97%, and 79%, respectively (P = .01). There were no differences in mentor effectiveness between groups. Mentored trainees were more likely to be satisfied with their programs (P = .01) and to report that faculty supported their professional aspirations (P = .001). Across all programs, fewer first-year trainees (59%) identified a mentor compared to PGY-2 through PGY-8 trainees (84%, P < .001).

CONCLUSIONS

A supported mentorship program is an effective way to create an educational environment that maximizes trainees' perceptions of mentorship and satisfaction with their training programs.

Cyclic hypobaric hypoxia improves markers of glucose metabolism in middle-aged men

Chronic hypoxia increases dependence on glucose in men and increases insulin sensitivity in men and women. Cyclic Variations in Altitude Conditioning (CVAC) is a novel technology that provides exposure to rapidly fluctuating cyclic hypobaric hypoxia (CHH).

PURPOSE

To test the hypothesis that markers of glucose metabolism would change with CVAC CHH, two groups of middle-aged men were exposed to 10 weeks (40 min/day, 3 day/week) of either CHH or sham (SH) sessions.

METHODS

CHH subjects (age: 48 ± 6, weight: 86 ± 12 kg, BMI: 27.1 ± 3, n=11) experienced cyclic pressures simulating altitudes ranging from sea level to 3048 m (week 1) and progressing to 6096 m (by week 5 through week 10). SH subjects (age: 50 ± 4, weight: 89 ± 15 kg, BMI: 27.5 ± 3, n=10) were exposed to slowly-fluctuating pressures up to 607 m (all subjects blinded to elevation). Physical function and blood markers of glucose metabolism were measured at baseline, 3, 6, and 10 weeks.

RESULTS

Two CHH subjects were dropped from analysis for failure to progress past 3048 m (CHH: n=9). Weight and physical activity remained stable for both groups. There was a group-by-time interaction in fasting glucose (CHH: 96 ± 9 to 91 ± 7 mg/dL, SH: 94 ± 7 to 97 ± 9 mg/dL, p<0.05). Reduction in plasma glucose response to oral glucose tolerance test [area under the curve] was greater in CHH compared to SH after 10 weeks of exposure (p<0.03). Neither group experienced changes in fasting insulin, insulin response during the OGTT, or changes in a timed walk test.

CONCLUSION

Ten weeks of CVAC CHH exposure improves markers of glucose metabolism in middle-aged men at risk for metabolic syndrome.