Diabetes mellitus and blood glucose variability increases the 30-day readmission rate after kidney transplantation

INTRODUCTION

Inpatient hyperglycemia is an established independent risk factor among several patient cohorts for hospital readmission. This has not been studied after kidney transplantation. Nearly one-third of patients who have undergone a kidney transplant reportedly experience 30-day readmission.

METHODS

Data on first-time solitary kidney transplantations were retrieved between September 2015 and December 2018. Information was linked to the electronic health records to determine diagnosis of diabetes mellitus and extract glucometric and insulin therapy data. Univariate logistic regression analysis and the XGBoost algorithm were used to predict 30-day readmission. We report the average performance of the models on the testing set on bootstrapped partitions of the data to ensure statistical significance.

RESULTS

The cohort included 1036 patients who received kidney transplantation; 224 (22%) experienced 30-day readmission. The machine learning algorithm was able to predict 30-day readmission with an average area under the receiver operator curve (AUC) of 78% with (76.1%, 79.9%) 95% confidence interval (CI). We observed statistically significant differences in the presence of pretransplant diabetes, inpatient-hyperglycemia, inpatient-hypoglycemia, minimum and maximum glucose values among those with higher 30-day readmission rates. The XGBoost model identified the index admission length of stay, presence of hyper- and hypoglycemia, the recipient and donor body mass index (BMI) values, presence of delayed graft function, and African American race as the most predictive risk factors of 30-day readmission. Additionally, significant variations in the therapeutic management of blood glucose by providers were observed.

CONCLUSIONS

Suboptimal glucose metrics during hospitalization after kidney transplantation are associated with an increased risk for 30-day hospital readmission. Optimizing hospital blood glucose management, a modifiable factor, after kidney transplantation may reduce the risk of 30-day readmission.

Who should see the patient? on deviations from preferred patient-provider assignments in hospitals

In various organizations including hospitals, individuals are not forced to follow specific assignments, and thus, deviations from preferred task assignments are common. This is due to the conventional wisdom that professionals should be given the flexibility to deviate from preferred assignments as needed. It is unclear, however, whether and when this conventional wisdom is true. We use evidence on the assignments of generalist and specialists to patients in our partner hospital (a children's hospital), and generate insights into whether and when hospital administrators should disallow such flexibility. We do so by identifying 73 top medical diagnoses and using detailed patient-level electronic medical record (EMR) data of more than 4,700 hospitalizations. In parallel, we conduct a survey of medical experts and utilized it to identify the preferred provider type that should have been assigned to each patient. Using these two sources of data, we examine the consequence of deviations from preferred provider assignments on three sets of performance measures: operational efficiency (measured by length of stay), quality of care (measured by 30-day readmissions and adverse events), and cost (measured by total charges). We find that deviating from preferred assignments is beneficial for task types (patients' diagnosis in our setting) that are either (a) well-defined (improving operational efficiency and costs), or (b) require high contact (improving costs and adverse events, though at the expense of lower operational efficiency). For other task types (e.g., highly complex or resource-intensive tasks), we observe that deviations are either detrimental or yield no tangible benefits, and thus, hospitals should try to eliminate them (e.g., by developing and enforcing assignment guidelines). To understand the causal mechanism behind our results, we make use of mediation analysis and find that utilizing advanced imaging (e.g., MRIs, CT scans, or nuclear radiology) plays an important role in how deviations impact performance outcomes. Our findings also provide evidence for a "no free lunch" theorem: while for some task types, deviations are beneficial for certain performance outcomes, they can simultaneously degrade performance in terms of other dimensions. To provide clear recommendations for hospital administrators, we also consider counterfactual scenarios corresponding to imposing the preferred assignments fully or partially, and perform cost-effectiveness analyses. Our results indicate that enforcing the preferred assignments either for all tasks or only for resource-intensive tasks is cost-effective, with the latter being the superior policy. Finally, by comparing deviations during weekdays and weekends, early shifts and late shifts, and high congestion and low congestion periods, our results shed light on some environmental conditions under which deviations occur more in practice.

Use of Imputation and Decision Modeling to Improve Diagnosis and Management of Patients at Risk for New-Onset Diabetes After Transplantation

BACKGROUND New-onset diabetes after transplantation (NODAT) is a complication of solid organ transplantation. We sought to determine the extent to which NODAT goes undiagnosed over the course of 1 year following transplantation, analyze missed or later-diagnosed cases of NODAT due to poor hemoglobin A1c (HbA1c) and fasting blood glucose (FBG) collection, and to estimate the impact that improved NODAT screening metrics may have on long-term outcomes. MATERIAL AND METHODS This was a retrospective study utilizing 3 datasets from a single center on kidney, liver, and heart transplantation patients. Retrospective analysis was supplemented with an imputation procedure to account for missing data and project outcomes under perfect information. In addition, the data were used to inform a simulation model used to estimate life expectancy and cost-effectiveness of a hypothetical intervention. RESULTS Estimates of NODAT incidence increased from 27% to 31% in kidney transplantation patients, from 31% to 40% in liver transplantation patients, and from 45% to 67% in heart transplantation patients, when HbA1c and FBG were assumed to be collected perfectly at all points. Perfect screening for kidney transplantation patients was cost-saving, while perfect screening for liver and heart transplantation patients was cost-effective at a willingness-to-pay threshold of $100 000 per life-year. CONCLUSIONS Improved collection of HbA1c and FBG is a cost-effective method for detecting many additional cases of NODAT within the first year alone. Additional research into both improved glucometric monitoring as well as effective strategies for mitigating NODAT risk will become increasingly important to improve health in this population.

Incidence, Risk Factors, and Trends for Postheart Transplantation Diabetes Mellitus

This retrospective study analyzed glycemic trends, incidence of post-transplant diabetes mellitus (PTDM) incidence and associated risk factors in a cohort of patients who underwent first-time heart transplantation (HT). Univariate analyses compared patient with and without pretransplant diabetes mellitus (DM). Multivariate regression analyses were conducted to determine association between PTDM and different risk factors. Finally, trends in glucometrics and other outcomes are described across follow-up time points. There were 152 patients who underwent HT between 2010 and 2015, 109 of whom had no pretransplant history of DM. PTDM incidence was 38% by the 1-year follow-up. Pretransplant body mass index (odds ratio [OR] 1.12, 95% confidence interval [CI] 1.01 to 1.23, p = 0.03), insulin use during the final 24 hours of inpatient stay (OR 4.26, 95% CI 1.72 to 10.56, p <0.01), mean inpatient glucose (OR 2.21, 95% CI 1.33 to 3.69, p <0.01), and mean glucose in the final 24 hours before discharge (OR 1.29, 95% CI 1.03 to 1.60, p = 0.03) were associated with increased odds of PTDM at 1 year. In patients on insulin before discharge, blood glucose values were significantly higher compared with those who were not (136 mg/dl vs 114 mg/dl at 1 to 3 months, 112 vs 100 at 4 to 6 months, 109 vs 98 at 8 to 12 months, all p <0.01). This analysis improves understanding of PTDM incidence, glucometric trends, and risk differences by DM status in the HT population. Similar to liver and kidney patients, inpatient glucometrics may be informative of PTDM risk in HT patients. Guidelines for this population should be developed to account for risk heterogeneity and need for differential management.

Comparison of post-transplantation diabetes mellitus incidence and risk factors between kidney and liver transplantation patients

Background

Most prior studies characterizing post-transplantation diabetes mellitus (PTDM) have been limited to single-cohort, single-organ studies. This retrospective study determined PTDM across organs by comparing incidence and risk factors among 346 liver and 407 kidney transplant recipients from a single center.

Methods

Univariate and multivariate regression-based analyses were conducted to determine association of various risk factors and PTDM in the two cohorts, as well as differences in glucometrics and insulin use across time points.

Results

There was a higher incidence of PTDM among liver versus kidney transplant recipients (30% vs. 19%) at 1-year post-transplant. Liver transplant recipients demonstrated a 337% higher odds association to PTDM (OR 3.37, 95% CI (1.38–8.25), p<0.01). 1-month FBG was higher in kidney patients (135 mg/dL vs 104 mg/dL; p < .01), while 1-month insulin use was higher in liver patients (61% vs 27%, p < .01). Age, BMI, insulin use, and inpatient FBG were also significantly associated with differential PTDM risk.

Conclusions

Kidney and liver transplant patients have different PTDM risk profiles, both in terms of absolute PTDM risk as well as time course of risk. Management of this population should better reflect risk heterogeneity to short-term need for insulin therapy and potentially long-term outcomes.

Interphysician Differences in Emergency Department Length of Stay

BACKGROUND

Emergency physicians differ in many ways with respect to practice. One area in which interphysician practice differences are not well characterized is emergency department (ED) length of stay (LOS).

OBJECTIVE

To describe how ED LOS differs among physicians.

METHODS

We performed a 3-year, five-ED retrospective study of non-fast-track visits evaluated primarily by physicians. We report each provider's observed LOS, as well as each provider's ratio of observed LOS/expected LOS (LOS_O/E); we determined expected LOS based on site average adjusted for the patient characteristics of age, gender, acuity, and disposition status, as well as the time characteristics of shift, day of week, season, and calendar year.

RESULTS

Three hundred twenty-seven thousand, seven hundred fifty-three visits seen by 92 physicians were eligible for analysis. For the five sites, the average shortest observed LOS was 151 min (range 106-184 min), and the average longest observed LOS was 232 min (range 196-270 min); the average difference was 81 min (range 69-90 min). For LOS_O/E, the average lowest LOS_O/E was 0.801 (range 0.702-0.887), and the average highest LOS_O/E was 1.210 (range 1.186-1.275); the average difference between the lowest LOS_O/E and the highest LOS_O/E was 0.409 (range 0.305-0.493).

CONCLUSION

There are significant differences in ED LOS at the level of the individual physician, even after accounting for multiple confounders. We found that the LOS_O/E for physicians with the lowest LOS_O/E at each site averaged approximately 20% less than predicted, and that the LOS_O/E for physicians with the highest LOS_O/E at each site averaged approximately 20% more than predicted.

The durability of operational improvements with rotational patient assignment

INTRODUCTION

Previous work has suggested that Emergency Department rotational patient assignment (a system in which patients are algorithmically assigned to physicians) is associated with immediate (first-year) improvements in operational metrics. We sought to determine if these improvements persisted over a longer follow-up period.

METHODS

Single-site, retrospective analysis focused on years 2-4 post-implementation (follow-up) of a rotational patient assignment system. We compared operational data for these years with previously published data from the last year of physician self-assignment and the first year of rotational patient assignment. We report data for patient characteristics, departmental characteristics and facility characteristics, as well as outcomes of length of stay (LOS), arrival to provider time (APT), and rate of patients who left before being seen (LBBS).

RESULTS

There were 140,673 patient visits during the five year period; 138,501 (98.7%) were eligible for analysis. LOS, APT, and LBBS during follow-up remained improved vs. physician self-assignment, with improvements similar to those noted in the first year of implementation. Compared with the last year of physician self-assignment, approximate yearly average improvements during follow-up were a decrease in median LOS of 18min (8% improvement), a decrease in median APT of 21min (54% improvement), and a decrease in LBBS of 0.69% (72% improvement).

CONCLUSION

In a single facility study, rotational patient assignment was associated with sustained operational improvements several years after implementation. These findings provide further evidence that rotational patient assignment is a viable strategy in front-end process redesign.

Emergency Department Holding Orders

BACKGROUND

Holding orders help transition admitted emergency department (ED) patients to hospital beds.

OBJECTIVE

To describe the effect of ED holding orders.

METHODS

We conducted a single-site retrospective study of ward admissions from the ED to the hospital internal medicine (HIM) service over 2 years. Patients were classified based on whether the ED did (group 1) or did not (group 2) write holding orders; group 1 was subdivided into patients sent to the floor with only ED holding orders (group 1A) vs. with subsequent HIM admission orders (group 1B). Outcomes were ED length of stay (LOS), time from decision to admit to ED departure (D→D), transfer to a higher level of care within 6 h (potential undertriage), and discharge from admission ward within 12 h (potential overtriage).

RESULTS

There were 9501 admissions: 6642 in group 1 (2369 in group 1A and 4273 in group 1B) and 2859 in group 2. Reductions in mean LOS between groups (with 95% confidence intervals [CIs] of the differences) were as follows: group 1 vs. 2: 44 min (39-49 min); group 1A vs. 1B, 48 min (43-53 min); group 1B vs. 2: 27 min (22-32 min); group 1A vs. 2: 75 min (69-81 min). Mean D→D was shorter in group 1A than 1B by 43 min (40-45 min). Holding orders were not associated with increases in potential undertriage or overtriage.

CONCLUSIONS

ED holding orders were associated with improved ED throughput, without evidence of undertriage or overtriage. This work supports the use of holding orders as a safe and effective means to improve ED patient flow.

Physician in Triage Versus Rotational Patient Assignment

BACKGROUND

Physician in triage and rotational patient assignment are different front-end processes that are designed to improve patient flow, but there are little or no data comparing them.

OBJECTIVE

To compare physician in triage with rotational patient assignment with respect to multiple emergency department (ED) operational metrics.

METHODS

Design-Retrospective cohort review. Patients-Patients seen on 23 days on which we utilized a physician in triage with those patients seen on 23 matched days when we utilized rotational patient assignment.

RESULTS

There were 1,869 visits during physician in triage and 1,906 visits during rotational patient assignment. In a simple comparison, rotational patient assignment was associated with a lower median length of stay (LOS) than physician in triage (219 min vs. 233 min; difference of 14 min; 95% confidence interval [CI] 5-27 min). In a multivariate linear regression incorporating multiple confounders, there was a nonsignificant reduction in the geometric mean LOS in rotational patient assignment vs. physician in triage (204 min vs. 217 min; reduction of 6.25%; 95% CI -3.6% to 15.2%). There were no significant differences between groups for left before being seen, left subsequent to being seen, early (within 72 h) returns, early returns with admission, or complaint ratio.

CONCLUSIONS

In a single-site study, there were no statistically significant differences in important ED operational metrics between a physician in triage model and a rotational patient assignment model after adjusting for confounders.

Characterization of Remitting and Relapsing Hyperglycemia in Post-Renal-Transplant Recipients

Background

Hyperglycemia following solid organ transplant is common among patients without pre-existing diabetes mellitus (DM). Post-transplant hyperglycemia can occur once or multiple times, which if continued, causes new-onset diabetes after transplantation (NODAT).

Objective

To study if the first and recurrent incidence of hyperglycemia are affected differently by immunosuppressive regimens, demographic and medical-related risk factors, and inpatient hyperglycemic conditions (i.e., an emphasis on the time course of post-transplant complications).

Methods

We conducted a retrospective analysis of 407 patients who underwent kidney transplantation at Mayo Clinic Arizona. Among these, there were 292 patients with no signs of DM prior to transplant. For this category of patients, we evaluated the impact of (1) immunosuppressive drugs (e.g., tacrolimus, sirolimus, and steroid), (2) demographic and medical-related risk factors, and (3) inpatient hyperglycemic conditions on the first and recurrent incidence of hyperglycemia in one year post-transplant. We employed two versions of Cox regression analyses: (1) a time-dependent model to analyze the recurrent cases of hyperglycemia and (2) a time-independent model to analyze the first incidence of hyperglycemia.

Results

Age (P = 0.018), HDL cholesterol (P = 0.010), and the average trough level of tacrolimus (P<0.0001) are significant risk factors associated with the first incidence of hyperglycemia, while age (P<0.0001), non-White race (P = 0.002), BMI (P = 0.002), HDL cholesterol (P = 0.003), uric acid (P = 0.012), and using steroid (P = 0.007) are the significant risk factors for the recurrent cases of hyperglycemia.

Discussion

This study draws attention to the importance of analyzing the risk factors associated with a disease (specially a chronic one) with respect to both its first and recurrent incidence, as well as carefully differentiating these two perspectives: a fact that is currently overlooked in the literature.

Emergency Department Rotational Patient Assignment

STUDY OBJECTIVE

We compare emergency department (ED) operational metrics obtained in the first year of a rotational patient assignment system (in which patients are assigned to physicians automatically according to an algorithm) with those obtained in the last year of a traditional physician self-assignment system (in which physicians assigned themselves to patients at physician discretion).

METHODS

This was a pre-post retrospective study of patients at a single ED with no financial incentives for physician productivity. Metrics of interest were length of stay; arrival-to-provider time; rates of left before being seen, left subsequent to being seen, early returns (within 72 hours), and early returns with admission; and complaint ratio.

RESULTS

We analyzed 23,514 visits in the last year of physician self-assignment and 24,112 visits in the first year of rotational patient assignment. Rotational patient assignment was associated with the following improvements (percentage change): median length of stay 232 to 207 minutes (11%), median arrival to provider time 39 to 22 minutes (44%), left before being seen 0.73% to 0.36% (51%), and complaint ratio 9.0/1,000 to 5.4/1,000 (40%). There were no changes in left subsequent to being seen, early returns, or early returns with admission.

CONCLUSION

In a single facility, the transition from physician self-assignment to rotational patient assignment was associated with improvement in a broad array of ED operational metrics. Rotational patient assignment may be a useful strategy in ED front-end process redesign.