Redesigned Care Delivery for Insulin-Requiring Diabetes in Pregnancy Improves Perinatal Glycemic Control While Reducing Neonatal Intensive Care Admissions,Length of Stay,and Costs

Abstract

Objective:

We sought to improve perinatal glycemic control and downstream neonatal outcomes through redesigned ambulatory management for women with insulin-requiring diabetes in pregnancy.

Methods:

To address gaps in perinatal glycemic management of women with insulin-requiring diabetes in pregnancy, redesigned care delivery (RCD) utilized integrated practice unit and minimally disruptive medicine concepts with incorporation of cellular-enabled glucose monitoring. Primary outcomes of RCD (N = 129) included hemoglobin A1c ([HbA1c], within RCD cohort), and gestational age (GA) at delivery, neonatal intensive care (NICU) admission, and NICU length of stay (LOS) compared with a preredesign care cohort (Pre-RCD; N = 122). Secondary outcomes included facility, payer reimbursement, and program costs. Generalized linear models assessed continuous variables while logistic regression methods assessed categorical outcomes.

Results:

Utilizing RCD, 92% of women with an initial HbA1c <6.5% maintained glycemic control until delivery, and 67.2% with an initial HbA1c ≥6.5% achieved delivery levels <6.5%. NICU admissions and GA-adjusted LOS decreased significantly [Pre-RCD vs. RCD: NICU admissions, 41.0% vs. 27.3%, p < 0.024; NICU LOS (95% confidence interval [CI]), 21.9 (17.1–26.6) vs. 14.6 (9.1–20.1), p = 0.045]. Every 10 days of redesigned management decreased mean NICU LOS by 1 day. Mean payer neonatal reimbursements decreased over $18,000 per delivery (p = 0.08) compared with implementation costs of $1,942 per delivery.

Conclusion:

Redesigned perinatal diabetes care with remote glucose monitoring demonstrated improved outcomes and value through downstream neonatal outcomes and lower payer costs. Therefore, subsequent dissemination and sustainability of similar programs' improved outcomes will likely require payer support.

Introduction

Diabetes affects 6.9% of U.S. pregnancies¹ and is associated with significant maternal self-management disease burden due to restrictive diets, intensive self-blood glucose monitoring, and frequent medication adjustments.^2,3 Frequent glucometer-to-log disparities, modifications of glucose log entries, and patients' minimization of the occurrence of hyperglycemia^4
–6 associated with traditional self-reported written glucose logs can all impact the safety and effectiveness of clinical decision making.

Despite effective medications and management guidelines for diabetes in pregnancy, failures to achieve desired maternal glycemic control culminate in costly adverse maternal and neonatal outcomes.^2,3 The economic burden of gestational diabetes mellitus has been estimated at 1.6 billion dollars,⁷ and avoidable complications of pregestational diabetes are estimated at 5.4 billion dollars annually.⁸ With their high self-care burden, problems of data quality critical to clinical management, and costly adverse outcomes despite available effective therapeutics, this population presents an opportunity for redesigned care delivery (RCD) to produce greater value (health outcomes per cost across time).^9

–12

Locally observed adverse maternal and neonatal outcomes, linked to inadequate maternal glycemic control, prompted the re-evaluation and redesign of our system of care for women with insulin-requiring diabetes in pregnancy. We focused care redesign and glycemic management upon women with insulin-requiring diabetes in pregnancy given their neonates' higher rates of neonatal intensive care (NICU) admissions and significant respiratory distress independent of gestational age (GA).¹³

In our earlier papers utilizing subpopulations from this improvement initiative, cellular-enabled glucometers were identified by patients as a better lifestyle fit,¹⁴ associated with improved glycemic control from care enrollment to delivery,¹⁵ and demonstrated the association between adherence to recommended testing and decreased odds of cesarean delivery, neonatal hypoglycemia, and large for GA neonates.¹⁶ Utilizing SQUIRE 2.0 criteria,¹⁷ we now compare our improvement initiative's associated maternal, neonatal, and fiscal outcomes following our redesigned ambulatory care deploying cellular-enabled glucometers specific to women with insulin-requiring diabetes in pregnancy.

Methods

This redesigned care improvement initiative was conducted at a tertiary care academic medical center's high-risk pregnancy clinic and children's hospital, and deemed not to be human subjects research (University of Iowa IRB #201509749).

Before redesigning care, we examined our existing perinatal diabetes management structures, processes,¹⁸ and domains of quality¹⁹ for gaps in effectively supporting maternal glycemic self-management. Opportunities to redesign our system of care with improved support of maternal glycemic control, and to thereby decrease downstream adverse neonatal outcomes became evident through morbidity and mortality conferences, discussions with faculty, residents, and nurses, as well as retelling of patient's stories.

Identified opportunities included: (1) increasing utilization of available multidisciplinary expertise; (2) establishing accountability for defining, measuring, and achieving meaningful outcomes through the application of integrated practice unit (IPU) care concepts^11,20; (3) enhancing standardization, communication, and flow of actionable data; and (4) shifting clinical decision making's reliance from traditional and lower data quality written glucose logs to utilization of more minimally disruptive²¹ and patient-centered^14,22 options offering greater automation of data capture and transmission.

RCD cohort

As detailed in Figure 1, beginning November 25, 2015, all women with insulin-requiring diabetes in pregnancy (diabetes mellitus Type 1, diabetes mellitus Type 2, and maturity-onset diabetes of the young or gestational diabetes mellitus A2) between 6 and 29 6/7 weeks gestation at perinatal diabetes care registration, and who intended to receive ongoing care, were offered cellular-enabled glucometers within the initiative. As multiple pregnancies increase prematurity and hypertensive disease risks in pregnancy, only the singleton outcomes of delivered pregnancies were included in the RCD cohort analysis. Women who opted not to participate and those ineligible for cellular-enabled glucometers due to late diabetes care registration (≥30 weeks gestation) received the same team care structure and standardized glycemic protocol management but relied upon standard glucometers and written logs.

FIG. 1.

Delineation of the perinatal diabetes program's redesigned care cohort (n = 129).

For comparison, a pre-RCD cohort (Fig. 2, “Pre-RCD,” n = 122) was derived utilizing an electronic data warehouse records search of women with insulin-requiring diabetes who both received prenatal care and delivered in the 18-month period immediately preceding redesigned care implementation (April 2014 to November 2015). For the final analysis, women with a “RCD” delivery were excluded from the “Pre-RCD” cohort.

FIG. 2.

Identification of the preredesigned care cohort (n = 122) of women with insulin-requiring diabetes in pregnancy.

Demographic, diagnosis, and clinical outcome variables were sourced directly from the organization's data warehouse utilizing discrete electronic medical record (EMR; EPIC, Verona, WI) inputs. Appropriate, large or small for GA status was calculated as previously described.²³ Corresponding maternal and neonatal length of stay (LOS), facility cost, and payer reimbursement data were sourced directly from the financial reporting system (ESPI, Enterprise Performance Solutions, Inc., Allscripts, Chicago, IL). Additional details of database validity are available in Supplementary Appendix SA1. All data were deidentified centrally before statistical analysis.

RCD resourcing

Acquisition costs of cellular-enabled glucometers, test strips, and salary for the perinatal diabetes program coordinator were internally funded. The coordinator position represented new programmatic support designated to perform new and different work created by redesigned care. Provisioned cellular-enabled glucometers (Telcare™, BioTelemetry, Inc., Concord, MA) are Food and Drug Administration-cleared, HIPAA-compliant, and automate uploading of glucose results to cloud-based patient and care team portals. Data transmission from cellular-enabled glucometers occurs without patients' requirements of mobile phones, smartphones, data plans, or Internet access, and the device provides text messaging to the patient. The physician champion's relative value unit (RVU) productivity target was reduced by 10% to support her administrative role.

RCD structure and process changes

The structural framework for RCD was based upon IPU and serious illness concepts,^11,20,24 where: (1) multidisciplinary care is centered around patients with a common disease process across the cycle of care; (2) teams leverage their diverse cognitive expertise; (3) management structures enhance colocation, communication, information flow, and timely data reviews; and (4) teams are accountable for defining and achieving meaningful outcomes. The interdisciplinary perinatal diabetes care team included a maternal/fetal medicine physician champion (J.I.A.), a medical endocrinology advanced practice provider and certified diabetes educator (R.F.), a health delivery redesign program champion (C.H.S.), the program coordinator (D.K.F.), an informatics reporting expert (J.M.K.), and a maternal/fetal medicine fellow (S.A.W.).

Team care delivery redesign actions included: (1) team role delineations; (2) standardization of protocols (screening, glucose test frequency, glycemic targets, medications, and dosing); (3) segregation of the insulin-requiring diabetes in pregnancy population away from the high-risk obstetrical clinics and into the perinatal diabetes program's specific care management; (4) interdisciplinary team reviews (weekly/biweekly) for all new patients, patients failing to achieve goals of care (fasting and/or postprandial values), emergency department or inpatient admissions, outcomes of deliveries, results of baseline to delivery hemoglobin A_1c (HbA1c), and NICU admissions from this population; and (5) quarterly tableau display comparisons of initiative to historic outcomes with formal reporting of program results to executive sponsors.

Our ambulatory cycle of care was process mapped, including gestational diabetes screening (in general obstetrics), disposition of referral requests, scheduling of antenatal visits with testing, inclusive of the postpartum visit following delivery. Team-based process mapping sessions were performed to evaluate and improve the safety, timeliness, efficacy, efficiency, equity, or patient centeredness¹⁹ of existing ambulatory care utilizing diverse ad hoc input from clinic administrators, nurses, residents, nutrition services, clinical information systems, and Lean engineering personnel. This input resulted in multiple process changes in screening, reporting, and scheduling access aimed at removing barriers to earlier diabetes care entry.

Compared with the use of written glucose logs, cellular-enabled glucometers were selected for automation of data to its portal to impact glucose data safety, flow efficiency, and data access. Eliminating manual glucose recording and reporting burden offered more “minimally disruptive medicine.”²¹ Cellular-enabled glucometers were provisioned at no cost to all participants to obviate potential social equity barriers to participation, such as payer reimbursement or digital health care access.^25,26

Utilization of the cellular-enabled glucometers' data portal and dashboards provided the care team with direct information flow and visual cueing assistance for timely identification of “outliers” to testing adherence or glucose results (tagged for meals and medications). Based upon those results, proactive supportive contact between scheduled follow-up was directed to those: (1) experiencing poor testing adherence, or (2) requiring interval insulin management changes based upon portal glucose results to glycemic targets.

To assist perinatal diabetes population management, daily EMR “pushed reporting” identified any inpatient admissions, emergency department or labor and delivery visits; and any nonobstetrical ambulatory clinic visits of redesigned care participants. Prior telephonic contact was replaced by patients' preference-driven contact options utilizing phone, text, and/or secure patient EMR portal.

Standardized glycemic management care protocols were implemented,^15,16 and consistent with the current American College of Obstetricians and Gynecologists and American Diabetes Association (ADA) guidelines.^2,3,27 Before RCD, HbA1c was not employed as a standard assessment of initial maternal glycemic control or to evaluate management effectiveness. RCD initiated the utilization of HbA1c assessments at care entry and delivery as standard practice with inadequate maternal glycemic control defined as HbA1c ≥6.5%.³ Individuals' disease and self-management educational needs were surveyed (Michigan Diabetes Knowledge Test²⁸ and Morisky Medication Adherence Survey²⁹) and addressed as part of the initiative.

RCD outcomes

Primary outcomes included maternal glycemic control (enrollment and delivery repeated measures of HbA1c, for the RCD cohort), and adverse neonatal events defined and measured by (1) preterm delivery (GA at delivery), (2) the occurrence of NICU admissions, and (3) NICU LOS. Secondary outcomes included maternal and newborn facility costs, payer reimbursements, and redesigned care's implementation costs ($USD/delivery).

Implementation costs per delivery represent program costs per delivery and the estimated cost of the provider RVU reduction per delivery where: (1) program costs per pregnancy were calculated from the coordinator's estimated time spent for this population (assigned proportionate paid salary and benefits), plus the acquisition costs of cellular-enabled glucometers and test strip vials dispensed for pregnancies; and (2) the estimated cost of the physician champion's RVU productivity target reduction calculated as target RVUs' reduced times mean maternal fetal medicine reimbursement per RVU.

Analysis

Descriptive statistics were calculated for all variables using frequencies (percent) or mean (±standard deviation). Outliers were checked for accuracy and possible data entry errors. Distributions of continuous variables were evaluated for normality assumptions by graphical observation. Collinearity among variables was examined using the Variance Inflation Factor method. To determine if there were associations between categorical variables and the study groups, Fisher's exact tests were used. Generalized linear models were used to assess differences between study groups for continuous variables, and logistic regression methods were used to assess associations with dichotomous outcomes.

Some models were adjusted for GA at delivery and length of time in the study. Some analyses using only the intervention group were adjusted for GA at first contact and/or length of enrollment in the study. Adjusted means (95% confidence intervals [95% CIs]) or odds ratios and 95% CI (ORs, 95% CI) were used to display the magnitude of differences between groups. As this was an exploratory improvement initiative, and as there is no consensus in the literature regarding whether and when corrections (e.g., Bonferroni) should be used for multiplicity,³⁰ we did not perform corrections for multiple comparisons to avoid the potential introduction of false negatives.³¹ This is so we could identify important trends to guide next steps that may otherwise have not been considered because of a lack of statistical significance.

The data analyses were generated using SAS software, Version 9.4 of the SAS System (Copyright © 2016 SAS Institute, Inc., Cary, NC). Statistical significance was set at alpha ≤0.05, but larger p-values are also presented for clinically meaningful differences.

Results

Of the 372 pregnant women with diabetes who planned ongoing care through the perinatal diabetes program, almost half (47.8%, n = 178) were eligible and offered cellular-enabled glucometers. Of those, 11.8% (21/178) declined (Fig. 1) and 2.5% (4/157) agreed but discontinued cellular-enabled glucometer use (all within 72 hours). In total, 83.4% (129/153) of women both continued cellular-enabled glucose monitoring and delivered singleton pregnancies to comprise the final RCD cohort. Reported findings are based on maternal and neonatal outcomes, associated facility costs, and adjudicated reimbursement data (censored after December 31, 2018) from these 129 singleton deliveries (December 4, 2015 to June 30, 2018) of women with insulin-requiring diabetes in pregnancy (Fig. 1).

Sociodemographic characteristics of the Pre-RCD and RCD cohorts are compared in Table 1. There were no significant differences in parity, race, body mass index, payer mix, GA at delivery, or route of delivery. Nearly half were insured by Medicaid. A greater proportion of women with type 2 diabetes were in the RCD cohort; however, significant diabetes type-specific data are missing from the Pre-RCD cohort. In addition, compared with the RCD initiative cohort, women who declined or withdrew from cellular-enable glucometer use (N = 25), were significantly more likely to have type 1 diabetes [RCD 26.4% (34/129) vs. Declined/Withdrew 72% (18/25), p < 0.0003]. While not a prespecified outcome, following process mapping and improvement, RCD was associated with a significant increase in mean days of prenatal care overall (130.9 vs. 158.1 days, p = 0.002) and for women with diabetes mellitus type 2 in particular (113.2 vs. 156.8 days, p = 0.001).

Table 1.

A Comparison of Sociodemographic Characteristics Between Preredesigned Care Delivery and Redesigned Care Delivery Cohorts (Total Women, N = 251)

	Pre-RCD women	RCD women	p^a
Participants, n (%)	122 (48.6)	129 (51.4)
Demographic variables
Age, mean (95% CI)^a	31.3 (30.2–32.4)	31.7 (30.7–32.8)	0.58
Race, n (%)			0.40
Black	15 (12.5)	24 (18.6)
Hispanic	9 (7.5)	14 (10.8)
Caucasian	86 (71.7)	81 (62.8)
Other	10 (8.3)	10 (7.8)
Missing data	2	None
Marital status n (%)			0.96
Married/partner	59 (54.6)	73(56.6)
Divorced/separated	8 (7.4)	9 (7.0)
Single	41 (38.0)	47 (36.4)
Missing data	14	0
Para, n (%)			0.69
0	41 (33.6)	39 (30.2)
1	34 (27.9)	45 (34.9)
2	25 (20.5)	25 (19.4)
>2	22 (18.0)	20 (15.5)
Diabetes type, n (%)			0.075
Type 1	11 (13.6)	34 (26.4)^b
Type 2	49 (60.5)	62 (48.1)^b	0.03^b
Gestational diabetes A2	21 (25.9)	31 (24.0)
Maturity-onset diabetes of young	0 (0.0)	2 (1.5)
Provider coding lacked specificity for above diabetes types	41 (33.6)	0
Prenatal body mass index (kg/m²), mean (95% CI)^a	37.1 (35.3–38.9)	36.6 (34.8–38.5)	0.75
Missing data	22	37
Payer mix, n (%)			0.70
Commercial	61 (50.0)	63 (48.8)
Medicaid	58 (47.5)	60 (46.5)
Other governmental	3 (2.5)	6 (4.7)
Missing data	0	0
Days of RCD enrollment, mean (95% CI)^a	NA	145.1 (134.6–155.6)	NA
Days of prenatal care, mean (95% CI)^a
Overall	130.9 (118.8–143.0)	158.1 (146.3–169.9)	0.002
Type 1	138.0 (95.8–180.2)	169.3 (145.3–193.2)	0.200
Type 2	113.2 (94.4–132.1)	156.8 (140.0–173.5)	0.001
Gestational diabetes A2	143.6 (116.0–171.2)	150.5 (127.8–173.2)	0.700
Maturity-onset diabetes of young	NA	128.5^c	NA
Provider coding lacked specificity for above diabetes types	41	0

p-Values for categorical data were generated using Fisher's Exact test; p-values and 95% CIs comparing means were generated from unadjusted general linear models, unless otherwise noted.

Statistically significant difference in the proportion of Diabetes Type 1 and Type 2 between pre-RCD and RCD.

Calculated 95% CI (−780 to 1037) not meaningful as based upon 2 values (57 and 200 days).

CI, confidence interval; NA, not applicable; RCD, redesigned care delivery.

Table 2 compares and details the impact of redesigned care on glycemic management maternal HbA1c levels obtained at care entry and delivery for the RCD cohort. With redesigned care, 92% of enrollees who entered with an HbA1c <6.5% maintained their controlled HbA1c throughout pregnancy. Over half of all women (56.3%) who entered redesigned care had an initial HbA1c ≥6.5%, and redesigned management was associated with nearly two-thirds (67.2%) of them achieving an HbA1c <6.5% at delivery. Despite declining mean HbA1c from entry to delivery, about one-third (32.8%) of women who entered redesigned care with poor glycemic control continued to have their HbA1c ≥6.5% at delivery. For those women who either declined (n = 21) or withdrew (n = 4) from cellular-enabled glucometer use, the mean % HbA1c (standard error [SE]) at entry was not significantly different from the RCD cohort [7.5% (0.4) vs. 7.2% (0.2), p > 0.05].

Table 2.

The Association of Maternal Metabolic Control as Assessed by HemoglobinA1_C Groups with NICU Admission Risk Following Redesigned Care for Insulin-Requiring Diabetes in Pregnancy

Enrollment HbA1c (mean, 95% CI)^a	Delivery HbA1c (mean, 95% CI)^a	n^b (%)	NICU admission, yes n (%)	NICU admission, no n (%)	Categorized HbA1c control and NICU risk OR (95% CI)^c
Controlled 5.7 (5.5–5.8)	Controlled 5.5 (5.4–5.6)	48 (40.3)	8 (25.0)	40 (46.0)	1.0
Controlled 5.8 (5.2–6.5)	Not controlled 7.1 (6.0–8.2)	4 (3.4)	1 (3.1)	3 (3.5)	1.7 (0.1–18.6)
Not controlled 8.2 (7.7–8.6)	Controlled 5.7 (5.6–5.8)	45 (37.8)	12 (37.5)	33 (37.9)	1.8 (0.7–5.0)
Not controlled 8.6 (7.9–9.2)	Not controlled 7.4 (6.8–8.0)	22 (18.5)	11 (34.4)	11 (12.6)	5.0 (1.6–15.7)
	Total	119	32	87

Controlled (C): HgbA1_C <6.5%; Not controlled (NC): HgbA1_C ≥6.5.

Mean and 95% CI were generated from unadjusted general linear models.

Ten of original 129 subjects are missing enrollment and/or delivery HgbA1c values.

ORs and 95% CIs were generated from unadjusted logistic models.

HbA1c, hemoglobin A1c; NICU, neonatal intensive care; OR, odds ratio.

After designating glycemic control at entry and delivery as either “controlled” (HbA1c <6.5%) or “not controlled” (HbA1c ≥6.5%), Table 2 provides risk estimates for NICU admission according to maternal entry and delivery HbA1c status specific to the RCD cohort. Compared with women who both entered and delivered with an HbA1c <6.5%, neonates of women with an HbA1c ≥6.5% at both entry and at delivery experienced the highest odds of NICU admissions (OR, 95% CI: 5.0, 1.6–15.7).

Regardless of HbA1c status at entry, adjusted for GA and days of redesigned care enrollment, neonates of women whose HbA1c was not controlled (vs. controlled) by delivery had significantly greater odds of NICU admission (not tabled; OR, 95% CI: 3.9, 1.4–10.7, p = 0.008). Conversely, after adjusting for both GA at delivery and for days of redesigned enrollment, maintaining or achieving a controlled HbA1c by delivery was associated with significantly reduced odds of NICU admissions (not tabled; OR, 95% CI: 0.35, 0.09–0.69, p = 0.008) compared with the presence of poor control (HbA1c ≥6.5%) at delivery.

Table 3 compares the two (Pre-RCD vs. RCD) cohorts' associated maternal and neonatal outcomes. Compared with the Pre-RCD cohort, there were no significant RCD-associated maternal reductions in the cesarean delivery rate, LOS, mean facilities costs, or payer reimbursements. Notably, the increase in days of prenatal care (Table 1) occurred without a shift in either mean GA at delivery or across GA groupings (Table 3).

Table 3.

Redesigned Care for Insulin-Requiring Perinatal Diabetes: A Comparison of Maternal and Neonatal Outcomes for Preredesigned Care Delivery and Redesigned Care Delivery Cohorts (N = 251)

	Pre-RCD	RCD	p^a
Participants, n (%)	122 (48.6)	129 (51.4)
Maternal outcomes
GA at delivery
Weeks at delivery, mean (95% CI)^a	37.5 (37.1–37.9)	37.6 (37.2–38.1)	0.70
Weeks at delivery, n (%)			0.29
≤33 6/7	7 (5.7)	8 (6.2)
34 0/7–36 6/7	29 (23.8)	19 (14.7)	0.072^b
37 0/7–38 6/7	38 (31.1)	50 (38.8)
≥39	48 (39.4)	52 (40.3)
Missing data	None	None
Delivery type, n (%)			0.31
Vaginal	48 (39.3)	60 (46.5)
Cesarean	74 (60.7)	69 (53.5)
Missing data	None	None
Maternal LOS, mean days (95% CI)^a	4.3 (3.4–5.3)	4.4 (3.4–5.3)	98
Missing data	0	0
Maternal facilities costs, mean dollars (95% CI)^a	7,828 (6,238–9,420)	9,512 (7,999–11,025)	0.13
Missing data	8	2
Maternal payer reimbursements, mean dollars (95% CI)^a	11,036 (9,483–12,589)	9,647 (8,167–11,124)	0.20
Missing data	8	2
Days of redesigned care enrollment, mean (95% CI)^a	NA	145.1 (134.6–155.6)	NA
Neonatal outcomes
Newborn size classifications n (%)			0.80
Appropriate for GA	74 (62.7)	86 (66.7)
Large for GA	35 (29.7)	35 (27.1)
Small for GA	9 (7.6)	7 (6.2)
Missing data	4	None
Newborn LOS, mean days (95% CI)^a	9.6 (6.8–12.5)	8.5 (5.8–11.2)	0.57
Missing data	3	1
Newborn NICU admittance, n (%)			0.024
Yes	50 (41.0)	35 (27.3)
No	72 (59.0)	93 (72.7)
Missing data	None	1^c
NICU LOS, mean days (95% CI)^a
Unadjusted	18.7 (13.1–24.3)	12.2 (5.5–18.9)	0.14
Adjusted for GA	21.9 (17.1, 26.6)	14.6 (9.1, 20.1)	0.045
Missing data	None	1^c
Newborn facilities costs, mean dollars (95% CI)^a	15,686 (9,482–21,890)	11,976 (6074–17,877)	0.39
Missing data	8	2
Newborn reimbursements, mean dollars (95% CI)^a	35,807 (21,212–50,402)	17,670 (3,787–31,552)	0.08

p-values for categorical data were generated using Fisher's Exact Test; p-values and 95% CIs comparing the mean were generated from unadjusted general linear models, unless otherwise noted.

Proportion of deliveries at 34 0/7–36 6/7 versus ≥39 weeks.

One neonate without any NICU admission, NICU LOS or cost data assignment secondary to intrapartum fetal death.

GA, gestational age; LOS, length of stay.

Compared with the neonates of the Pre-RCD cohort, the RCD cohort had a significantly lower percentage of neonates requiring NICU admission [41.0% (50/122) vs. 27.3% (35/128), p = 0.024]. Within the RCD cohort and according to NICU admission status (results not tabled), there were no differences in the mean GA at enrollment [17.0 weeks (95% CI: 14.3–19.8) vs. 16.9 weeks (95% CI: 15.2–18.6), p = 0.92] or mean days of care [134.6 (114.4–154.8) vs. 149.6 (95% CI: 137.2–162.0), p = 0.21]. Compared with Pre-RCD, the RCD cohort's mean (95% CI) NICU LOS, adjusted for the categorical GAs shown, was significantly reduced [Pre-RC, 21.9 (17.1–26.6) vs. RC, 14.6 (9.1–20.1), p = 0.045] (Table 3).

Additional analyses (not tabled) identified that the mean NICU LOS for GA 33 to ≥36 weeks declined significantly [pre-RC vs. RC, 16.3 (95% CI: 11.3–21.3) vs. 6.2 days (95% CI: 0.0–13.2), p = 0.023]. With NICU LOS as the dependent variable and enrollment as the independent variable, each day of redesigned care enrollment was associated with a 0.1-day reduction (p = 0.008) in NICU LOS. After adjusting for GA at enrollment, each day of redesigned enrollment was associated with a 0.3-day reduction in NICU LOS (p = 0.016).

The Pre-RCD to RCD cohorts' mean neonatal facility costs per delivery (Table 3) declined modestly (pre-RCD: $15,686 vs. RCD $11,976, p = 0.39). In contrast, the mean payer neonatal reimbursements declined substantially by over $18,000/delivery (p = 0.08). With estimated redesigned program costs of $1,167/delivery and provider champion costs of $775/delivery, the calculated program implementation costs/delivery were $1,942/delivery.

In the Pre-RCD cohort, 19.7% (24/122) of women had entered care at ≥30 weeks. A subset analysis adjusted for GA at delivery and according to Pre-RCD care entry (< vs. ≥30 weeks) showed that neither: (1) the rate of NICU admissions (37.8% vs. 54.2%, p = 0.17); (2) NICU LOS [mean (SE)] 28.1 (3.0) vs. 27.6 (5.1); or (3) neonatal reimbursements [(mean, 95% CI) $ 76,520 (47,322–105,718) vs. $78,767 (30,920–126,614)] differed significantly.

Discussion

We redesigned our care structure, processes, and glucose-monitoring technology for women with insulin-requiring diabetes in pregnancy with the goal of better supporting maternal glycemic self-management and minimizing downstream adverse maternal and neonatal outcomes. Redesigned processes were associated with improved access to care as measured by mean duration of prenatal care overall and for women with Type 2 diabetes in particular. Redesigned care incorporating cellular-enabled glucometers delivered improved maternal glycemic control with reduced adverse neonatal outcomes of NICU admissions and LOS at lower neonatal costs.

Disease burden induces poor adherence, wasted resources, and poor outcomes.²¹ Women with diabetes in pregnancy have previously identified the use of cellular-enabled glucometers to be a better lifestyle fit,¹⁴ and for women with type 2 and gestational diabetes utilizing cellular-enabled glucometers every 10% increase in adherence to testing recommendations decreases the odds of cesarean delivery, neonatal hypoglycemia, and large-for-gestational-age neonates by 15%–20%.¹⁶ Compared with standard glucometer and patient logs, direct uploads of patient data by remote monitoring technologies offer reduced disease burden with improved accuracy and safety of data for medical decision making.^4
–6

Critically, our redesigned care employed a series of processes, aided by a data portal, dashboards, and pushed reports to fully leverage remotely gathered patient data and foster population “management by exception.” Visual cueing of data within dashboards first assisted in the early identification of outliers to goals, which triggered the team's focused prioritization and proactive outreach to those with unmet testing or glycemic goals between scheduled contact. The maternal improvements in HbA1c and the reduced NICU admissions and LOS are reflective of the availability of timely, accurate data and ability to assist outliers to desired glycemic targets. Paired with Epic's “pushed reporting,” unplanned or unreported care utilization requiring acute follow-up could be identified.

Importantly, multidisciplinary reviews of all maternal and neonatal outcomes were engineered into the IPU structure and schedule to engage diverse expertise, promote communication, and accountability. Although not prespecified outcomes, the observed improvement in days of prenatal care and improved coding specificity for diabetes type following RCD (Table 1, “missing data”) likely reflect the benefits of addressing quality of care processes through mapping and standardization.

This current report extends our prior findings of HbA1c improvements utilizing cellular-enabled glucometers in women with insulin-requiring diabetes in pregnancy.¹⁵ Utilizing a different comparison cohort of women using standard glucometers (who either declined [n = 13] or were ineligible (GA ≥30 weeks, [n = 32] for cellular-enabled glucometers) versus a much smaller sample (n = 71) from the RCD quality improvement initiative, women with poor glucose control (HbA1c >6.5%) had significantly lower and greater improvement of HbA1c by delivery.¹⁵

In our current report utilizing cellular-enabled glucometers, nearly all women (92%) who entered RCD with adequate glycemic control remained in desired control; and, importantly, over two-thirds of women with inadequate glycemic control at entry successfully achieved glycemic control. As new findings, we observed that neonates of women with insulin-requiring diabetes in pregnancy whose HbA1c was not controlled by delivery had nearly four-fold greater odds of NICU admission. Conversely, the odds of an NICU admission were 65% less likely for neonates of women who successfully maintained or achieved desired glycemic control by delivery. Our demonstration of HbA1c ≥6.5% as an effective marker for NICU admission risks affirms prior research³ relating HbA1c to neonatal outcomes, despite physiologic decreases in HbA1c during early and late pregnancy.^32,33

With the absence of pre-RCD HbA1c data for comparison, Pre-RCD versus RCD NICU admissions and LOS served as surrogates for poor glycemic control and final arbiters of improved outcomes following RCD. We demonstrated a significant RCD-associated reduction in NICU admissions and a GA-adjusted difference in NICU LOS of over 7 days with lower payer costs trending at nearly $18,000 per newborn. Our findings of reduced hyperglycemia and differences in neonatal outcomes and costs contrast to both a meta-analysis that demonstrated minimal glycemic improvement,³⁴ with remote monitoring, and a randomized trial of remote monitoring in gestational diabetes, which found no differences in neonatal and cost outcomes.³⁵

We believe that our system-of-care approach, explicitly addressing structure, processes, and resourcing, and our population selection influenced this difference. As over 50% of our redesigned care cohort of women with insulin-requiring diabetes in pregnancy exhibited significant and poorly controlled diabetes at enrollment, very likely our demonstration of benefits to NICU admissions, GA-adjusted NICU LOS, and payer costs reflects our underlying decision to deploy cellular-enabled glucometers for a population at higher risk of significant hyperglycemia and neonatal complications compared with a gestational diabetes population.

In support of our findings of improvement in glycemic control, neonatal outcomes and their relevance to high-risk populations, the multisite (n = 31), randomized, open-label, CONCEPTT trial of self-management in pregnant women with type 1 diabetes³⁶ demonstrated improved HbA1c and NICU admissions for continuous glucose monitor users compared with capillary glucose monitoring. Likewise, our observed mean 1.1-day improvement in newborn hospital LOS, although not statistically significant, is similar to their finding of a 1-day improvement. However, additionally, we demonstrated reduced NICU LOS and payer costs.

We found that every 1 day of RCD was associated with a 0.1-day reduction in NICU LOS. Although RCD did increase the days of prenatal care, we believe the reported differences in NICU admissions and LOS stem from improved glycemic control under care. Technologies such as cellular-enabled glucometers, which provide more frequent, reliable, and actionable data, achieve better glycemic control than standard glucometers and paper-based log reporting. With better data, increasing the time in care provides more opportunities for coaching and treatment modifications to improve glycemic control to target and impact neonatal outcomes.

Our findings which support this include that: (1) regardless of HbA1c status at entry, adjusted for GA and days of redesigned care, neonates of women whose HbA1c was not controlled by delivery had significantly greater odds of NICU admission, and (2) within the RCD group, days of care did not determine which neonates were admitted to the NICU. This is further supported by CONCEPTT trial findings,³⁶ where all participants were enrolled before 14 weeks, but continuous glucose monitor users achieved the reduction in NICU admissions.

Technologies utilized to collect and transmit data continue to transition over time, from glucometers and paper logs, to cellular enable glucometers, to cloud based continuous glucose monitors. Broadening the impact of a new technology to meaningfully improve health outcomes rests upon several critical points: (1) how transmitted data is utilized by the patient and care team in a proactive and engaging manner, (2) the care structures and processes which support that system, (3) the selection of a population likely to benefit from that technology's application, and (4) the presence of factors which support or negate a technology's acceptance and utilization across differing populations.

A strength of our initiative was the provision of patients' access to an enhanced technology which automated transmission of their data to a care team and was deployed without patient costs while functioning across both urban and rural environs, without requirements for daily internet access, smart phone ownership, cellular data plans, and without regard to financial status or insurer. This is important because maximizing population benefits from a remote monitoring technology requires acceptance by those at risk while their broad participation can be adversely impacted by social equity barriers, such as financial capacity, payer reimbursement, or digital access.^25,26

Unfortunately, for many populations, differences in broadband access track along social, economic, and geographic divisions and for many negate the applicability of technologies, which require facile internet access. Nearly half of our patients utilized Medicaid where the financial eligibility requirement (133% of poverty level) represents an income under $21,000 for a family of two. In addition, nearly 20% of our state's population is considered rural. While the average rate for states' broadband adoption (2018) is 84%, households earning <$20,000 have a broadband adoption rate of only 62%.³⁷ Similar to other disparities in health outcomes, the lowest broadband adoption rates are associated with the lowest median incomes, highest shares of rural communities, and the highest shares of communities of color.³⁷

Further impediments to health technology can exist due to payer contracts and benefit design. During the RCD initiative, Medicaid contractors would not (and still do not) provide coverage in pregnancy for nonstandard glucose monitors and we remain unable to obtain Medicaid contractor authorization for continuous glucose monitors in pregnancy. Therefore, despite the market availability of continuous glucose monitors, provisioned cellular-enabled glucometers' ability to obviate the accessibility barriers described highlights the ongoing relevance of this work for women with diabetes in pregnancy, and in particularly those of vulnerable and under-resourced populations.

While our single-site findings are robust and promising, our initiative has design limitations related to “before and after” comparisons of existing management outcomes to redesigned care outcomes. The retrospective cohort analysis limits attribution of observations to associations, and not causality.

Lacking a randomized trial design, comparisons were subject to the potential limitations of: (1) the Pre-RCD patient population's issues of cohort balance, (2) potential for recruitment bias resulting from convenience and voluntary enrollment (or declination) for cellular-enabled glucometer use, and (3) the potential introduction of bias through time course changes in referral patterns, scheduling or treatment. Unlike RCD cellular-enabled glucometer recipients, the Pre-RCD cohort contained women who entered care ≥30 weeks. However, our analyses were reassuring for a lack of impact. Adjusted for GA at delivery, there were no significant differences within the Pre-RCD group's outcomes for NICU admissions, NICU LOS or neonatal reimbursements by care entry before or after 30 weeks gestation.

Likewise, individuals who declined or withdrew from RCD cellular-enabled glucometer use could introduce bias if their exclusion removed a group with significantly different (i.e., worse) baseline glycemic control. We believe such bias was unlikely as the mean HbA1c at care entry did not differ significantly for the decline/withdrew group in comparison to cellular-enabled RCD participants.

While coding specificity for the type of insulin-requiring diabetes following RCD was high; the lack of Pre-RCD EMR coding specificity for a significant proportion of the Pre-RCD group limits reliable comparison of pre-RCD to RCD balance by type of diabetes. While we can only speculate on women's motives for declining the enrollment for cellular-enabled glucometers, there was significantly higher (p < 0.0003) representation of women with type 1 diabetes (N = 18/25) in the declined/withdrew group compared with the RCD group (N = 34/129). Taken together, differences in Pre-RCD coding and cellular-enabled glucometer acceptance by women with type 1 diabetes could influence the difference in diabetes type between the Pre-RCD and RCD groups.

Undetected treatment time course bias, such as introduction of late preterm maternal betamethasone use, could impact NICU admissions and LOS outcomes. However, the institution of betamethasone therapy excluded women with type 1 or type 2 diabetes mellitus and did not change NICU admission risk or median LOS.³⁸ Further dispelling the likelihood of significant time course effects, the division of neonatology faculty was consulted with our results and neither identified a contemporaneous reduction in NICU general population LOS or NICU changes responsible for the LOS shift specific to neonates of women with insulin-requiring diabetes.

Moreover, several mutually supportive findings favor redesigned care as responsible for the observed improvements. These include: (1) longer duration of prenatal diabetes care after process mapping, (2) high rates of glycemic control achieved at delivery corresponded with reductions in NICU admission risk, (3) the decline from pre to post redesigned care NICU admissions and mean NICU LOS (adjusted for GA), and (4) the associated 1-day decline in NICU LOS for every 10 days of maternal enrollment.

It is important to note that this initiative was not intended to determine the upper GA for enrollment or minimum management time course required to obtain meaningful benefit from cellular-enabled glucometer usage. As noted, we believe the increasing days of care following RCD, across all groups and significantly for women with type 2 diabetes in particular, reflects improvements made through process mapping to enhance the timeliness and access to care. Despite those improvements, many women remained ineligible to receive cellular-enabled glucometers because of their late referral and registration for specialized diabetes care. Given the association of HbA1c improvement on NICU admission risk, and the impact of duration of enrollment on NICU LOS, examining and reducing upstream barriers to late referral or registration for diabetes care represents an opportunity to extend the neonatal benefits of RCD across a greater population of women with insulin-requiring diabetes in pregnancy.

Single-site financial comparisons may have limited generalizability because compensation, benefits, supplies, and overheads vary across institutions. However, any lower facility costs observed are encouraging as our unionized labor costs averaged 2% growth each year (2017 vs. 2016, 2018 vs. 2017) and medical supplies averaged 4% and 2% increases, respectively. Single-site payer reimbursement generalizations are limited by system-specific contractual payment agreements. However, our observed decreases in neonatal reimbursement per delivery are more likely due to changes in NICU utilization and LOS than contractual payment reductions as payer reimbursements (weighted for this population's specific payer mix) rose 0.54% per year during the same time frame.

Demonstrating statistical significance for our lower payer neonatal reimbursements per delivery would have required a sample size of 129 in each group (80% power, p < 0.05). In support of our findings, the magnitude of our nearly $18,000/delivery reduction in payer reimbursements for insulin-requiring diabetes in pregnancy compares favorably to 12-month cost differences of $14,600/infant attributed to longer LOS and higher unscheduled NICU visits.^39,40 Therefore, although lacking a statistical difference, as operational “decisions based on the costs, benefits and likelihoods of all potential consequences always beat those made based solely on statistical significance,”⁴¹ the magnitude of our reimbursement savings relative to implementation costs should be of interest to payers.

Conclusions

Diabetes in pregnancy remains an ongoing public health concern for women despite available effective treatment modalities. Our RCD for women with insulin-requiring diabetes in pregnancy incorporated IPU and process of care changes, including cellular-enabled glucometers. Redesigned care demonstrated significant improvements in maternal glycemic control, and meaningful improvements in downstream NICU admissions and LOS at reduced payer costs. This work has important implications for integrating maternal and neonatal outcomes and costs when evaluating the value^9,10,12 of perinatal diabetes care, redesigning programs and the incorporation of technology to reduce disease burden and bridge health disparities.

As cost savings and reimbursements are powerful drivers to either changing or continuing the status quo, realizing the demonstrated neonatal benefits through subsequent dissemination, adoption, and sustainability for similar redesign programs will likely require aligning support from accrued payer savings. Given the universal commonalities of patient disease burden, reliance on glucose data for safe and effective care management, and adverse outcomes associated with poor maternal glycemic control, opportunities to effectively redesign care delivery for women with insulin-requiring diabetes in pregnancy exist beyond our institution.

Footnotes

Acknowledgments

The authors wish to acknowledge the support shown by Scott Turner, MA, former executive director of the University of Iowa Stead Family Children's Hospital. The organizational skills of Ms. Elesa W. Wedemeyer, MAT, MHA were instrumental in launching this initiative. The role of Ms. Rhonda M. Fruhling, MS, ARNP, CDE as a diabetes care expert and perinatal diabetes team member is recognized. S.A.W. is funded as a research trainee of the Fraternal Order of Eagles Research Center at the University of Iowa (T32DK112751).

Author Disclosure Statement

C.H.S., S.A.W., and J.I.A. declare that no competing financial interests exist. D.K.F., J.M.K., and L.M.R. declares that no financial interests exist.

Funding Information

This initiative was, and continues to be, funded by the University of Iowa Health Care.

Supplementary Material

Supplementary Appendix SA1

References

Deputy

, Kim

, Conrey

, Bullard

. Prevalence and changes in preexisting diabetes and gestational diabetes among women who had a live birth—United States, 2012–2016. MMWR Morb Mortal Wkly Rep, 2018; 67:1201–1207.

American College of Obstetricians and Gynecologists' Committee on Practice Bulletins. Gestational Diabetes Mellitus. ACOG Practice Bulletin No. 190 Obslet Gynecol, 2018; 131(2):e49–e63.

American Diabetes Association. 14. Management of diabetes in pregnancy: Standards of medical care in diabetes-2019. Diabetes Care, 2019; 42:S165–S172.

Cosson

, Baz

, Gary

, et al. Poor reliability and poor adherence to self-monitoring of blood glucose are common in women with gestational diabetes mellitus and may be associated with poor pregnancy outcomes. Diabetes Care, 2017; 40:1181–1186.

Given

, O'Kane

, Bunting

, Coates

. Comparing patient-generated blood glucose diary records with meter memory in diabetes: A systematic review. Diabet Med, 2013; 30:901–913.

Kendrick

, Wilson

, Elder

, Smith

. Reliability of reporting of self-monitoring of blood glucose in pregnant women. J Obstet Gynecol Neonatal Nurs, 2005; 34:329–334.

Dall

, Yang

, Gillespie

, et al. The economic burden of elevated blood glucose levels in 2017: Diagnosed and undiagnosed diabetes, gestational diabetes mellitus, and prediabetes. Diabetes Care, 2019; 42:1661–1668.

Peterson

, Grosse

, Li

, et al. Preventable health and cost burden of adverse birth outcomes associated with pregestational diabetes in the United States. Am J Obstet Gynecol, 2015; 212:74.e71–e79.

Berwick

, Nolan

, Whittington

. The triple aim: Care, health, and cost. Health Aff (Millwood), 2008; 27:759–769.

10.

Porter

. What is value in health care?. N Engl J Med, 2010; 363:2477–2481.

11.

Porter

, Lee

. The strategy that will fix healthcare. Harvard business review. Boston, MA: Harvard Business Publishing,, 2013:50–70.

12.

Sudhof

, Shah

. In pursuit of value-based maternity care. Obstet Gynecol, 2019; 133:541–551.

13.

Becquet

, El Khabbaz

, Alberti

, et al. Insulin treatment of maternal diabetes mellitus and respiratory outcome in late-preterm and term singletons. BMJ Open, 2015; 5:e008192.

14.

Pierick

, Wernimont

, Boevers

, et al. Enhanced patient satisfaction with cellular-enabled glucose meters in pregnancy. J Diabetes Sci Technol, 2018; 12:1086–1087.

15.

Wernimont

, Sheng

, Fleener

, Summers

, Syrop

, Andrews

. Cellular-enabled glucometers and maternal glucose control: A quality improvement initiative. J Diabetes Sci Technol, 2019; 14:77–82.

16.

Wernimont

, Sheng

, Tymkowicz

, et al. Adherence to self-glucose monitoring recommendations and perinatal outcomes in pregnancies complicated by diabetes mellitus. Am J Obstet Gynecol MFM, 2019; 1:100031.

17.

Ogrinc

, Davies

, Goodman

, Batalden

, Davidoff

, Stevens

. SQUIRE 2.0 (Standards for QUality Improvement Reporting Excellence): Revised publication guidelines from a detailed consensus process. BMJ Qual Saf, 2016; 25:986–992.

18.

Donabedian

. Evaluating the quality of medical care. 1966. Milbank Q, 2005; 83:691–729.

19.

Crossing the quality chasm: A new health system for the 21st century. Institute of Medicine Committee on Quality of Health Care in America. Washington, DC: National Academies Press, 2001.

20.

Porter

, Teisberg

. Redefining health care: Creating value-based competition on results. Boston, MA: Harvard Business School Press, 2006.

21.

May

, Montori

, Mair

. We need minimally disruptive medicine. BMJ, 2009; 339:b2803.

22.

Hirst

, Mackillop

, Loerup

, et al. Acceptability and user satisfaction of a smartphone-based, interactive blood glucose management system in women with gestational diabetes mellitus. J Diabetes Sci Technol, 2015; 9:111–115.

23.

Fenton

, Kim

. A systematic review and meta-analysis to revise the Fenton growth chart for preterm infants. BMC Pediatr, 2013; 13:59.

24.

Van Houtven

, Hastings

, Colon-Emeric

. A path to high-quality team-based care for people with serious illness. Health Aff (Millwood), 2019; 38:934–940.

25.

Perrin

. Digital gap between rural and nonrural America persists. 2019. Available at: https://www.pewresearch.org/fact-tank/2019/05/31/digital-gap-between-rural-and-nonrural-america-persists/ Accessed August 30, 2019 .

26.

Anderson

, Kumar

. Digital divide persists even as lower-income Americans make gains in tech adoption. 2019. Available at: https://www.pewresearch.org/fact-tank/2019/05/07/digital-divide-persists-even-as-lower-income-americans-make-gains-in-tech-adoption/ Accessed August 30, 2019 .

27.

American College of Obstetricians and Gynecologists' Committee on Practice

Bulletins

, Pregestational Diabetes

Mellitus

. ACOG Practice Bulletin No. 201. Obstet Gynecol, 2018; 132:e228–e248.

28.

Fitzgerald

, Funnell

, Anderson

, Nwankwo

, Stansfield

, Piatt

. Validation of the revised brief diabetes knowledge test (DKT2). Diabetes Educ, 2016; 42:178–187.

29.

Morisky

, Ang

, Krousel-Wood

, Ward

. Predictive validity of a medication adherence measure in an outpatient setting. J Clin Hypertens (Greenwich), 2008; 10:348–354.

30.

Streiner

, Norman

. Correction for multiple testing: Is there a resolution?. Chest, 2011; 140:16–18.

31.

Rothman

. No adjustments are needed for multiple comparisons. Epidemiology, 1990; 1:43–46.

32.

Mosca

, Paleari

, Dalfra

, et al. Reference intervals for hemoglobin A1c in pregnant women: Data from an Italian multicenter study. Clin Chem, 2006; 52:1138–1143.

33.

Nielsen

, Ekbom

, Damm

, et al. HbA1c levels are significantly lower in early and late pregnancy. Diabetes Care, 2004; 27:1200–1201.

34.

Ming

, Mackillop

, Farmer

, et al. Telemedicine technologies for diabetes in pregnancy: A systematic review and meta-analysis. J Med Internet Res, 2016; 18:e290.

35.

Mackillop

, Hirst

, Bartlett

, et al. Comparing the efficacy of a mobile phone-based blood glucose management system with standard clinic care in women with gestational diabetes: Randomized controlled trial. JMIR Mhealth Uhealth, 2018; 6:e71.

36.

Feig

, Donovan

, Corcoy

, et al. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): A multicentre international randomised controlled trial. Lancet, 2017; 390:2347–2359.

37.

Tomer

, Fishbane

, Siefer

, Callahan

. Digital prosperity: How broadband can deliver health and equity to all communities. 2020. Available at: https://www.brookings.edu/research/digital-prosperity-how-broadband-can-deliver-health-and-equity-to-all-communities/ Accessed June 22, 2020 .

38.

Gyamfi-Bannerman

, Thom

, Blackwell

, et al. Antenatal betamethasone for women at risk for late preterm delivery. N Engl J Med, 2016; 374:1311–1320.

39.

Adams

, Nishimura

, Merritt

, Melvin

. Costs of poor birth outcomes among privately insured. J Health Care Finance, 2003; 29:11–27.

40.

Law

, McCoy

, Lynen

, et al. Costs of newborn care following complications during pregnancy and delivery. Matern Child Health J, 2015; 19:2081–2088.

41.

Amrhein

, Greenland

, McShane

. Scientists rise up against statistical significance. Nature, 2019; 567:305–307.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB