Cellular-Enabled Glucometers and Maternal Glucose Control: A Quality Improvement Initiative

Abstract

Background:

Management of diabetes in pregnancy is burdensome due to self-glucose monitoring, recording, and reporting demands. Cellular-enabled glucometers provide real-time transmission of glucose values independent of internet access and cell phone data plans. We describe a quality improvement (QI) intervention that introduced cellular-enabled glucometers for use during pregnancies complicated by diabetes.

Methods:

Our aim was to improve maternal glucose control in a cohort of insulin-requiring pregnant women enrolled in a telemedicine diabetes program. During initial establishment of a QI program, women were offered cellular-enabled glucometers but could elect to keep their standard meter. The primary outcome evaluated was glycosylated hemoglobin A1c (HbA1c) at delivery.

Results:

Baseline characteristics including initial HbA1c were similar between women using a standard glucometer (n = 45) and those using a cellular-enabled glucometer (n = 72). Women who used a cellular-enabled glucometer had a lower HbA1c at delivery compared to those using a standard glucometer (5.8% vs 6.3%, P = .03). This improvement was particularly notable for women with poor glucose control (defined as HbA1c >6.5%) at initial obstetric visit. Women with poor glucose control who used a cellular-enabled glucose monitor had significantly lower HbA1c at delivery (6.0% vs 6.8%, P = .03) and greater change from initial visit compared to those using a standard glucometer (−2.6% vs −1.4%, P = .02). No statistically significant differences were detected in tracked neonatal outcomes.

Conclusion:

For pregnancies complicated by insulin-requiring diabetes, use of cellular-enabled glucometers as part of a perinatal diabetes program improves glucose control at delivery with timely transmission of accurate values throughout gestation.

Diabetes in pregnancy carries significant maternal and neonatal risks, and optimal management requires frequent self-glucose monitoring.^1-3 Self-monitoring of blood glucose can be burdensome for patients with patients citing time constraints and fear of abnormal results as barriers to effective self-management in pregnancy.^4,5 While many clinics rely on self-reported blood glucose values provided by patients during clinic visits or weekly phone calls, this methodology does not always result in reliable values for clinical decision making or offer timely interventions to patients. Prior studies comparing patient reported values and glucometer logs in pregnancy have shown that up to 40% of self-reported glucose values are modified⁶ and these inaccuracies may impact clinical decisions.⁷

Telemedicine has the potential to improve management of pregnancies complicated by diabetes,^8,9 though supporting data is limited. HIPAA compliant cellular-enabled glucometers enable real-time transmission of blood glucose values to care teams independent of patient internet access and cell phone data plans. These meters provide accurate and timely glucose values and are not dependent on patient manual entry, eliminating one source of reporting error. Additionally, since these glucometers rely on cellular service independent of a patient’s individual plan or internet access, there are fewer barriers to glucose reporting.

As part of a quality improvement (QI) initiative to improve maternal glucose control in pregnancy, we introduced use of cellular-enabled glucometers. We aimed to reduce burdens of patient reporting of glucose values, ensuring that timely and accurate information was relayed to a multidisciplinary perinatal diabetes team. Locally, we often questioned the reliability of self-reported glucose values provided on patient supplied logs and the potential adverse impact on clinical decision making and outcomes. We hypothesized that pregnant women with insulin-requiring diabetes who used cellular-enabled glucose monitoring would have better glucose control at delivery due to timely transmission of reliable and accurate glucose values to care teams. Utilizing SQUIRE 2.0 guidelines,¹⁰ we report the results of this QI intervention: introducing cellular-enabled glucometers to improve maternal glucose control as defined by HbA1c at delivery.

Methods

We performed a cohort study of pregnant women requiring insulin enrolled in a telemedicine diabetes program as part of a QI initiative. A key objective of this program was to improve maternal glucose control prior to delivery as defined by HbA1c at time of delivery. As a QI initiative, this project was reviewed by the University of Iowa Institutional Review Board (IRB) and was deemed not human subjects research (IRB no. 201509749).

During an 18-month transitional phase, all women at less than 29 6/7 weeks’ gestation and requiring insulin were offered cellular-enabled glucometers (Telcare™, Biotelemetry Inc, Concord MA, USA), or they could elect to keep their standard meter upon enrollment (Supplemental Figure 1). We did not offer a cellular-enabled glucometer if patients presented to care after 29 weeks’ gestation. Thus two cohorts were examined: those using standard glucometers with paper-log-based reporting and those using cellular-enabled glucometers (Figure 1). Women using standard glucometers recorded glucose values on supplied log sheets and were contacted weekly by a diabetes care nurse to report values. Women using cellular-enabled glucometers had values automatically uploaded to a secure, web-based care team portal in real time, and values were reviewed by the diabetes care team at least weekly.

Figure 1.

Study sample of women enrolled in perinatal diabetes program.

Both groups received similar diabetes care in pregnancy. Irrespective of meter type and in accordance with ADA recommendations,¹ we recommended 4 blood glucose tests daily (fasting and 1 hour after meals) for women with gestational and type 2 diabetes and 7 blood glucose tests daily (fasting, before and 1 hour after meals) for women with type 1 diabetes. Insulin was administered via multiple daily injections to all women with type 2 and gestational diabetes. For women with type 1 diabetes, half of women in each group used an insulin pump to administer insulin (50% in the standard group vs 47% in the cellular enabled group) while the remaining administered insulin via multiple daily injections.

We included in all women meeting criteria who presented after November 2015 and delivered at our institution before July 2017. Except for the use of the cellular based glucometers, both groups received care from the same multidisciplinary diabetes care team with weekly communication with a diabetes care nurse and regular clinic follow-up with a maternal-fetal medicine care team.

Women were excluded for the following criteria: delivery at an outside facility, spontaneous or therapeutic abortion, or lack of HbA1c at delivery. Baseline HbA1c was obtained upon entry to prenatal care for women with type 1 and type 2 diabetes and for women with risk factors for gestational diabetes.² Women without an initial HbA1c were included in this study, contributing to delivery HbA1c and neonatal outcomes; however, they were excluded in pairwise analyses.

The primary outcome evaluated was change in HbA1c from entry to care to delivery. An a priori power analysis determined that a sample of 21 patients per group would be required to detect a 1% difference in change in HbA1c between groups from baseline to delivery using a two-sided independent samples t-test with a type I error rate of 5% and a power of 80%. This power calculation assumed a 0.5% decline in HbA1c from baseline to delivery in the standard glucometer group and a standard deviation of 1.1% across both groups as done in CONCEPPT trial.¹¹ Our sample size is larger than calculated to facilitate a preplanned subgroup analysis for women with poor glucose control at initial prenatal care visit (defined as HbA1c >6.5% by American Diabetes Association [ADA] guidelines).¹ All results were analyzed by chi-square, Fisher’s exact, or t-test as indicated. P < .05 was considered statistically significant.

Secondary outcomes were HbA1c at delivery, neonatal hypoglycemia, large for gestational age, and shoulder dystocia. Hypoglycemia was defined as a neonate with any blood glucose less than 47 mg/dL per hospital protocol. Large for gestational age was determined by birthweight greater than the 95th percentile for expected gestational age.¹² Patients were diagnosed with type 1 and type 2 diabetes prior to pregnancy. Our institution practices a two-step approach to diagnosis of gestational diabetes. Gestational diabetes was defined by elevation of a 3-hour glucose tolerance test in accordance with ACOG guidelines.¹³ Additionally, according to convention at the time, any diabetes first diagnosed during pregnancy was classified as gestational diabetes. Preeclampsia was defined by 2 blood pressures greater than 140 mmHg systolic and 90 mmHg diastolic and either 24 hours urine protein > 300 mg, a urine protein to creatinine ratio >0.3, or evidence of end organ damage.¹⁴ Gestational hypertension was defined as new onset hypertension in pregnancy.¹⁴ Chronic hypertension was defined as hypertension prior to 20 weeks’ gestation.¹⁵

Results

Baseline characteristics between the two groups are shown in Table 1. Of women in the standard group, 29% declined use of the cellular-enabled glucometer. The remaining were not offered, typically due to gestational age greater than 29 weeks with initial program contact. Four women withdrew from use of the cellular-enabled glucometer during pregnancy (Figure 1): one preferred her former meter and three opted to not use any meter in pregnancy. They are all included in the cellular-enabled glucometer analysis. While both groups shared a similar percentage of women with type 1 diabetes, the cellular-enabled glucometer group had more women with type 2 diabetes while the standard group included more women with gestational diabetes.

Table 1.

Baseline Characteristics of All Included Pregnant Patients.

	Standard glucometer (%)	Cellular-enabled glucometer (%)	P
Number of patients	45	72
• Telcare declined	13/45 (29%)	X
• Telcare not offered	32/45 (71%)	X
Age (years)	31 ± 4.7	32 ± 6.2	.572
BMI-kg/m2 at initial OB visit	35 ± 9.1	36 ± 8.4	.884
Nulliparous %	17/45 (38%)	18/72 (25%)	.153
Type of diabetes			.018‡
• Type 1 diabetes	14/45 (31%)	19/72 (26%)
• Type 2 diabetes	14/45 (31%)	40/72 (56%)
• GDMA2	17/45 (38%)	13/72 (18%)
HbA1c at initial OB visit	7.4 ± 1.81a	7.6 ± 1.84b	.962
Private insurance	27/45 (60%)	29/72 (40%)	.057
Chronic HTN	13/45 (29%)	24/72 (33%)	.685
Preeclampsia	11/45 (24%)	16/72 (22%)	.824
gHTN	4/45 (9%)	4/72 (6%)	.482
GA at delivery	37.19 ± 2.22	37.33 ± 2.59	.768
Vaginal delivery	26/45 (58%)	31/71c (44%)	.182
Cesarean delivery	19/45 (42%)	40/71c (56%)

Data are presented as mean ± SD for continuous variables, number (%) for categorical variables. P values were calculated for t-tests for continuous data, and Fisher’s exact tests for categorical data, unless otherwise noted.

^aMissing data for 13 cases.

^bMissing data for 6 cases.

^c1 participant with IUFD not included in the denominator.

^‡P value for chi-square test.

The two groups were similar in terms of age, BMI, parity, and coexistent chronic hypertension. There were no significant differences in gestational age at delivery, mode of delivery, or rates of preeclampsia or gestational hypertension (Table 1). Both groups had similar glucose control at initiation to program enrollment with HbA1c at initial OB visit of 7.4 ± 1.8% in the standard glucometer group and 7.6 ± 1.8% in the cellular-enabled glucometer group.

We found that women using a standard glucometer and those using a cellular-enabled glucometer both had a decrease in HbA1c over the course of pregnancy (Table 2). Average change in HbA1c from initial prenatal visit to delivery was not found to differ by more than 1% between groups. However, women in the cellular-enabled group overall achieved a lower HbA1c at time of delivery compared to those using the standard glucometer (5.8 ± 0.7% vs 6.3 ± 1.3%, P = .03). We found no statistically significant differences in the secondary outcomes of neonatal hypoglycemia, large for gestational age infants and shoulder dystocia at time of delivery (Table 2).

Table 2.

Outcomes for All Included Pregnant Patients.

	Standard glucometer	Cellular-enabled glucometer	P
Delivery HbA1c	6.3 ± 1.33	5.8 ± 0.71	.03e
Change HbA1c (initial-delivery HbA1c)	1.1 ± 1.3a	1.8 ± 1.75b	.07e
Neonatal outcome (%)
Hypoglycemia	24/45 (53%)	42/71f (59%)	.568d
Large for gestational age	11/45 (25%)	23/72 (32%)	.411d
Shoulder dystocia	6/25c (24%)	2/31c (7%)	.121d

^aMissing data for 13 cases.

^bMissing data for 6 cases.

^cDenominator is total vaginal deliveries.

^dP for Fisher’s exact test.

^eP for t-test.

^f1 participant with IUFD not included in the denominator.

We hypothesized that the impact of the cellular-enabled glucometer may be greatest for women with poor glucose control (HbA1c >6.5%) at initial obstetric visit. We found that women with poor glucose control in both groups did not differ significantly in any baseline characteristic (Table 3). The initial HbA1c in the standard group was 8.4% compared to 8.6% in the cellular-enabled group (Table 3). These are well above ADA preconception targets of 6.5%. At the time of delivery, women using the cellular-enabled glucometer had an average HbA1c of 6.0% compared to an average HbA1c of 6.8% for those women using a standard glucometer (Table 4). Further, the average decrease of HbA1c from initial visit to delivery was significantly greater for women using the cellular-enabled glucometer (−2.6 ± 1.7%) compared to those using a standard glucometer (−1.4 ± 1.4%).

Table 3.

Baseline Characteristics Women With Poor Initial Glucose Control Based on Initial HbA1c>6.5%.

	Standard glucometer (%)	Cellular-enabled glucometer (%)	P
Number of patients	22	43
Telcare declined	10/22 (46%)	X
Telcare not offered	12/22 (55%)	X
Age-year	31 ± 4.7	31 ± 6.2	.987
BMI-kg/m2 at NOB?	35 ± 9.9	35 ± 7.4	.810
Nulliparous %	9/22 (41%)	9/43 (21%)	.142
Type of diabetes
Type 1 diabetes	10/22 (45%)	18/43 (42%)
Type 2 diabetes	9/22 (41%)	25/43 (58%)
GODMA2	3/22 (14%)	0/43 (0%)
HbA1C at initial OB visit	8.4 ± 1.74	8.6 ± 1.53	.723
Private insurance	13/22 (59%)	15/43 (35%)	.071
Chronic HTN	7/22 (32%)	13/43 (30%)	1.00
Preeclampsia	7/22 (32%)	8/43 (19%)	.351
gHTN	1/22 (4.5%)	3/43 (7%)	1.00
GA at delivery	36.63 ± 2.62	37.47 ± 2.14	.168
Vaginal delivery	14/22 (64%)	15/42 (36%)	.062
Cesarean delivery	8/22 (36%)	27/42 (64%)

Data are presented as mean ± SD for continuous variables, number (%) for categorical variables. P values were calculated for t-tests for continuous data, and Fisher’s exact tests for categorical data, unless otherwise noted.

Table 4.

Primary Outcome for Women With Poor Initial Glucose Control Based on Initial HbA1c >6.5%.

	Standard glucometer	Cellular-enabled glucometer	P
Delivery HbA1c	6.8 ± 1.54	6.0 ± 0.72	.028
Change HbA1c (initial-delivery HbA1c)	1.4 ± 1.37	2.6 ± 1.65	.022
Neonatal outcome (%)
Hypoglycemia	11/22 (50%)	31/43 (72%)	.103
Large for gestational age	4/22 (18%)	16/43 (37%)	.159
Shoulder dystocia	4/13a (31%)	1/15a (7%)	.153

^aDenominator is total vaginal deliveries.

Discussion

As part of a QI initiative, we show that use of a cellular-enabled glucometer is associated with improved maternal glucose control in pregnancies complicated by diabetes requiring insulin. This improvement is especially notable for women entering pregnancy with poor glucose control. Upon enrollment in our program, all insulin adjustments were managed by the same multidisciplinary team irrespective of meter type. Use of a cellular-enabled glucometer in the context of a perinatal diabetes program improved glucose control. We believe that this is due to the timely transmission of accurate and actionable glucose values to the care team. We have previously shown that pregnant women with diabetes report significantly more satisfaction with a cellular-enabled glucometer compared to a standard glucometer especially in terms of how the equipment functions and how well it fits into their lifestyle.¹⁶

We used HbA1c as a marker of maternal glycemic control in pregnancy and differences between initial and delivery HbA1c were compared for most women in both groups. The use of HbA1c has been questioned as a perinatal outcome because it decreases in pregnancy due to increased red blood cell turnover.¹⁷ When HbA1c has been studied in nondiabetic populations, the upper range of normal decreases from 5.5% before pregnancy to 5.0% in the third trimester.¹⁸ HbA1c decreased by more than this expected amount for both groups of women enrolled in our perinatal diabetes program, suggesting that management with a perinatal diabetes team can improve glucose control in pregnancy. While the decrease in HbA1c from initial visit to delivery did not differ by more than 1% between women using the standard compared to the cellular-enabled glucometer in our full population, it is notable that the decrease was statistically significant in our population of pregnant women entering pregnancy with poor glucose control. We attribute this detection difference to the fact that women who entered pregnancy with HbA1c less than 6.5% are less likely to have further large decreases in HbA1c over the course of pregnancy. Most likely, initial poor glucose control allowed better detection of HbA1c changes. Therefore, the potential impact of this technology may be highest in women entering pregnancy with poor glucose control.

Our initiative is unique by only including women with more severe forms of diabetes. 49% of women using the standard glucometer and 60% of women using the cellular-enabled glucometer had HbA1c greater than 6.5% at enrollment. A further marker of disease severity is only inclusion of women requiring use of insulin in pregnancy. At the time of initial enrollment into our perinatal diabetes program, oral medications were still frequently used in pregnancy; thus women who had progressed to insulin typically had more severe disease. Recent recommendations advise use of insulin as a first line medication for diabetes when pharmacologic treatment is needed.^2,3

The severity of diabetes in our study allowed improvements in glucose control to be detected over the course of pregnancy. A recent randomized controlled trial of a smartphone based app (where patients manually entered glucose values) for management of pregnancies complicated by gestational diabetes did not show improvement in glucose control in pregnancy with the use of the mobile technology compared to standard care.¹⁹ However, with the average HbA1c at the time of trial enrollment at 5.4% for both groups, any impact on glucose control may have been difficult to measure. Thus, future studies of telemedicine tools in pregnancy may find the biggest impact when directed toward women with poor glucose control.

The major limitation of our study is selection bias and lack of randomization to meter use. While our two groups are similar, patient choice in choosing to continue with a standard glucometer or transition to a cellular-enabled glucometer may be impacted by patient comfort with current monitoring systems and may be a reflection of increased experience with diabetes management. Additionally two-thirds of our standard group were not offered a meter due to our program policy to not initiate cellular-enabled glucometer use after 29 weeks’ gestation. This led to more women with gestational diabetes in our standard meter group compared to our cellular-enabled group. Women who were comfortable with their standard meter and those with gestational diabetes requiring insulin after 29 weeks may have had less severe disease than those diagnosed earlier in pregnancy and electing use of a cellular-enabled meter. If so, the impact of our cellular-enabled glucometer is even greater than measured. Since our study was completed at a midwestern academic medical center where the majority of patients were referred from outside providers, the results may not be generalizable to other settings. Finally, this report does not specifically address cost of care related to cellular-enabled glucometer use, though our unpublished work has shown significantly less nursing time is required to obtain glucose values with use of a cellular-enabled glucometer.

Conclusions

As part of a QI initiative, we show that use of a cellular-enabled glucometer improved maternal glucose control in pregnancies complicated by diabetes requiring insulin and that its impact was greatest in women with poor glucose control early in pregnancy. This study was not powered to detect differences in neonatal outcomes; thus, larger studies will be needed to detect these important outcomes. The cellular-enabled glucometer allowed for timely transmission of accurate glucose values resulting in better management of diabetes in our perinatal program. Given the findings in this study, especially for women with poor glucose control, we have now widely implemented routine use of the cellular-enabled glucometer in our perinatal diabetes program.