Telemedicine in the treatment of gestational diabetes: pregnancy outcomes and maternal satisfaction.

PURPOSE The treatment of gestational diabetes requires several outpatient visits, from diagnosis until delivery, to prevent maternal and fetal complications associated to hyperglycemia. In literature there is poor evidence about the telemedicine systems superiority in improving pregnancy outcomes, in women with gestational diabetes. The aim of the study is to evaluate the maternal and fetal outcomes and the degree of satisfaction in gestational diabetes treatment, through exclusive telemedicine versus outpatient follow-up. METHODS 62 consecutive women with gestational diabetes were recruited by the Diabetology Unit of Ferrara: 29 randomized to a weekly remote control (telemedicine group); 33 checked in presence every two or three weeks (conventional group). To assess satisfaction with diabetes care, 58 women replied postpartum to the modi�ed Oxford Maternity Diabetes Treatment Satisfaction Questionnaire. RESULTS No statistically signi�cant differences were found in most of the obstetric and neonatal clinical parameters evaluated in both groups. The analysis of the questionnaire answers showed favorable scores in all areas explored. Telemedicine follow-up makes women feel more controlled (P= 0.007) and �ts better to their lifestyle (P = 0.001). It also emerged that almost all women treated with telemedicine would recommend this method to a relative or a friend. CONCLUSION Telemedicine follow-up proved to be safe both on metabolic control and pregnancy outcomes; furthermore, it signi�cantly decreases outpatient visits and increases women satisfaction. Studying the impact of telemedicine is also necessary, considering the di�culties associated with the Sars-COV-2 pandemic. In the future, it is desirable to expand the cohort of patients.


Introduction
Gestational Diabetes Mellitus (GDM) is de ned according to the American Diabetes Association, as "diabetes diagnosed in the second or third trimester of pregnancy that is not clearly manifest diabetes before gestation". In many cases the disease regresses after childbirth, so women have little time available to know and accept their condition [1]. The progressive increase in maternal hyperglycemia between 24 and 28 weeks of pregnancy is associated with short and long-term maternal-fetal risks [2]. Maternal complications include preeclampsia and caesarean section. Moreover, these women continue to check their blood sugar after childbirth at least every 3 years, due to the increased risk of developing type 2 diabetes, as the onset of gestational diabetes is often indicative of pancreatic beta cell dysfunction [1].
Fetal complications include an increased risk of macrosomia, neonatal hypoglycemia, hyperbilirubinemia, shoulder dystocia, and birth trauma. Furthermore, fetal exposure to maternal gestational diabetes contributes to the development of abnormalities in glucose metabolism and obesity in childhood and adulthood, regardless of genetic predisposition [3]. It seems that neonatal adiposity and fetal hyperinsulinemia, both higher if maternal hyperglycemia is constant, are mediators of childhood body fat [4]. Page 3/20 The frequency of obstetric checks increased in diabetic women. The treatment of gestational diabetes primarily involves changes in lifestyle (adequate diet and moderate physical activity) and, to evaluate the effectiveness of these treatments, daily blood sugar control is necessary. The blood glucose values measured are conventionally recorded on glycemic diaries, shown at each visit. So, it is necessary to carry out various checks to assess the adequacy of treatment and to adjust the therapy with insulin if necessary. The high number of prenatal checks affects compliance and the maternal psyche [5,6].
Furthermore, a percentage of women nd di cult to achieve their blood glucose goals. Several factors may be involved, such as personal beliefs about own health, poor understanding of the importance of good blood glucose control, inability to stick to a challenging regimen of blood glucose measurements, up to seven times a day, and the need to adjust insulin dosage [7].
Telemedicine can potentially reduce the number of visits and improve the quality of life of women, without increasing the occurrence of adverse neonatal and maternal outcomes [8].
Telemedicine can be de ned as a subgroup of telehealth that "uses communication networks for the provision of health services and medical extension from one geographic location to another, primarily to address challenges such as uneven distribution and shortage of infrastructural and human resources" [9].
Telemedicine has several effects, including overcoming the need for physical co-presence, improving the perception of always having a doctor available. A dimension often evaluated in relation to telemedicine is compliance, understood as the willingness to follow medical prescriptions (insulin doses or the number of glycemic measurements). Another relevant dimension is satisfaction with treatment and service, a multidimensional element that includes emotional and cognitive assessments. The quality of communication between patient and clinician is a key factor in uencing it. Personal recording of health data allows for an improvement in patient empowerment. When it comes to empowered women with GDM it means that "the patient should have a clear understanding of the disease, its pathogenesis, and its short-and long-term consequences for mother and child" [10].
There is insu cient evidence in the literature to a rm the superiority of telemedicine systems in the treatment of women with gestational diabetes. High-quality research is needed to determine the effectiveness and satisfaction of women and professionals [11].
Following the introduction of telemedicine in diabetic services, this study has two endpoints: to evaluate the pregnancy outcomes (primary endpoint) and to estimate the degree of satisfaction in the treatment of gestational diabetes (secondary endpoint).

Material And Methods
We performed a closed observational cohort study in a series of women affected by GDM, comparing a group followed in telemedicine and a group evaluated in presence. The study was performed in Ferrara (Italy) from February 2018 until August 2019. group" or the "conventional group". For the "conventional group'' the clinical follow-up took place at the Complex Operative Unit of Territorial Diabetology of Ferrara, with outpatients visits every two or three weeks, until the end of the pregnancy. For the "telemedicine group" there was only one medical examination in presence on the enrolment day, also including an interview with a dietician. Most of the women delivered at Operative Unit of Gynecology and Obstetrics of the Sant 'Anna University Hospital in Ferrara, following the established protocols.
Between the 24th and 28th gestational week the diagnosis of gestational diabetes was made using an oral load curve with 75 g of glucose (Oral Glucose Tolerance Test -OGTT), in adult women without preexisting diabetes. The "telemedicine group" women were provided with a glucometer, to perform the daily four-point glycemic self-check (fasting and one hour after the three main meals), automatically transferred to a virtual cloud through an application downloaded on their smartphone. Every week the same diabetologist checked all "telemedicine group" women's values on the telematic platform, verifying the achievement of the glycemic targets (fasting < 90 mg / dl, one hour after a meal < 130 mg / dl). In case of achievement of the targets, the diabetologist carried out one telemedicine visit per month; in case of persistent exceeding off the target values, he prescribed insulin therapy and the woman switched to a conventional follow-up.
Finally, in October 2020 women were contacted by telephone to answer the Questions and Responses to the Oxford Maternity Diabetes Treatment Satisfaction Questionnaire (OMDTSQ) by Hirst et al [12], modi ed ad hoc and translated into Italian language (Table 1). Table 1 Questions and Answers to the modi ed-Oxford Maternity Diabetes Treatment Satisfaction Questionnaire (OMDTSQ) modi ed. The rst nine questions were translated into Italian from the original OMDTSQ by Hirst et al [12], with quantitative answers to be assigned a score on a Likert-type scale (from + 3 to -3); the last four questions have qualitative answers, assigned by the authors. For each of the two groups, the following demographic variables were evaluated: age, parity, twin pregnancy, pre-pregnancy BMI (Body Mass Index), HbA1c (glycated hemoglobin) at diagnosis, gestational week at diagnosis of diabetes, native Italian language. The obstetric variables considered are weight gain at the end of pregnancy, disease of pregnancy (gestational hypertension, preeclampsia, IntraUterine Growth Restriction-IUGR, cholestasis), induction of labor and mode of induction (mechanical, prostaglandins, oxytocin and combined methods), time of delivery and mode of delivery (spontaneous delivery, caesarean section or vacuum extractor), postpartum blood loss, cases of shoulder dystocia.
The neonatal outcomes evaluated were birthweight, Large for Gestational Age (LGA) infants, macrosomal infants, hypoglycemia at birth, admission to Neonatal Intensive Care Unit (NICU), respiratory distress, hyperbilirubinemia and malformations.
As a secondary endpoint, the women satisfaction on diabetes care was investigated by telephone interview. In the rst part of the survey, they had to assess their agreement with the nine statements of the OMDTSQ (on general satisfaction with diabetes care, on the perception the relationship with the diabetes team and satisfaction with the technology used) giving a score on a seven-point Likert-type scale: from + Page 7/20 qualitative judgment on the number of visits performed, on their adherence to glycemic controls, on the performance of postpartum OGTT; nally, only the "telemedicine group" was asked if they would recommend this method to friends or relatives with gestational diabetes ( Table 1).
The data relating to glycemic control were retrieved from the computer databases used by the diabetes service: by a virtual cloud for the telemedicine group and by the values recorded by the glucometer for the conventional group. The maternal and neonatal clinical data derive from the medical records of the deliveries that occurred at the study center. Only in two cases of the telemedicine group, the letters of discharge of the birth that took place elsewhere were recovered. The questionnaire was submitted to women contacted randomly by telephone, by the same investigator.
To avoid selection bias, the two groups were selected randomly. To overcome language comprehension problems, the questionnaire was always administered by the same operator. Furthermore, it was not possible to avoid the medical surveillance bias (intrinsic to the nature of our study) as the telemedicine group was always controlled by the same diabetologist, while the conventional group was controlled by several diabetologists. The sample was chosen arbitrarily of the size of 30 women for each group.

Statistical Analysis
The T test was used to compare the continuous variables of the two groups; for the analysis of the observed frequencies the Chi-square test was used and, for the analysis of the frequencies of smaller samples, the Fisher test. The questionnaire was evaluated by comparing the average of the scores obtained for each response between the two groups, using the Mann-Whitney test for independent samples. A P value < 0.05 was set to calculate the static signi cance, calculated by MedCalc Software Ltd.

Results
From February 2018 to August 2019 among all pregnant women positive for the OGTT between the 24th and 28th week of gestation, 73 women were randomly selected, nine did not attend the visits and were excluded. 64 women were therefore enrolled as follows: 31 in telemedicine group, 33 in the conventional group. Two followed in telemedicine group suspended the follow-up for transfer to another hospital. Among the 62 total women followed up to childbirth (29 in telemedicine vs 33 in conventional), only 58 participated in the subsequent telephone survey (Fig. 1).
The two cohorts are homogeneous if we consider the demographic characteristics (p > 0.05), as shown in Table 2; in both groups, in fact, most of the sample is under the age of 35, is nulliparous with single pregnancy, overweight (pre-pregnancy BMI mean is 25 kg / m 2 ) and has HbA1c mean normal values at the beginning of pregnancy. In the conventional group, ve women speak a native language other than Italian. The use of telemedicine has made it possible to achieve the desired glycemic targets; in fact, as regards the diabetic outcomes, the glycemic compensation in both groups did not differ and no statistically signi cant differences were found in the need to resort to insulin treatment (P > 0.05). Regarding maternal outcomes (Table 2), there are no statistically signi cant differences between the two groups, neither in pregnancy pathologies and obstetric emergencies, nor in the modality and time of delivery (P > 0.05). The blood losses of three women are unknown. Table 2 Results of the demographic characteristics and outcome of pregnancy of the population. The results of the group followed with telemedicine and the group followed in a traditional way are compared; the values are shown as mean, standard deviation (SD) and median or number and relative percentage. The statistical signi cance analysis conducted for each variable compared when possible is also shown.  Table 3 shows the neonatal outcomes: there are no differences between the birthweights, rates of macrosomial infants and cases of hypoglycemia (P > 0.05). Hospitalization in NICU was higher for the telemedicine group (10 vs 2 infants, P = 0.01) and there was a greater number of cases of respiratory distress (7 vs 1 infants, P = 0.025). There were no differences in the other conditions investigated. Table 3 Results concerning neonatal outcomes. The results of the group followed with telemedicine and the group followed in a traditional way are compared; the values are shown as mean, standard deviation (SD) and median or number and relative percentage. The statistical signi cance analysis conducted for each variable compared when possible is also shown. e Types of malformations found (more than one per patient) in the telemedicine group: pyloric stenosis, macrocephaly and varus metatarsus, heart muscle interventricular defect, short lingual frenum; in the conventional group: short lingual frenum and hypospadias, bilateral syndactyly of the feet, buried penis, undescended testicle.
*Fisher's test The answers to the questionnaire are reported in Fig. 2. For the rst nine questions, most of the women gave positive scores (+ 2 and + 3), except for the question n.3 ("I am satis ed with my knowledge on gestational diabetes ") and the question n.9 (" This blood glucose monitoring has adapted to my lifestyle "), in which the scores are more distributed towards lower values. Only in the conventional group the maximum negative score (-3) was recorded, especially in the sphere that concerns the relationship with the team. It should be noted that for most respondents, the number of visits was adequate and there was good compliance with the execution of the measurements. At least 30% did not perform the postpartum OGTT.
For telemedicine group, 96% of respondents said they would recommend it to relatives and friends.
In the comparison between the averages of the scores obtained (Table 4), it was noted that women in telemedicine group reached the highest values in all the answers, with statistical signi cance in the statements: "I felt well controlled by the diabetes team" (P = 0.007) and "This blood glucose monitor has adapted to my lifestyle" (P = 0.001). Table 4 Comparison between the answers to the questionnaire obtained in the two groups, one followed with telemedicine, the other with the conventional method. Questions n.1-9 have scores from + 3 to -3: to compare the qualitative answers, the average of the scores obtained in each group was made. Questions n. 10-12 have qualitative answers, for which a value was assigned to each possible answer and then the average of the scores obtained in each group was made. Question n.10 value 1 = few, value 2 = right, value 3 = too many; question n.11 value 1 = always, value 2 = sometimes, value3 = almost never, value 4 = never; question n.12 value 1 = yes, value 2 = no. The P value shown has been calculated from the two scores obtained.

Discussion
To evaluate the clinical impact of telemedicine on the treatment of gestational diabetes, the comparison between telemedicine and conventional methods revealed that there were no signi cant differences in clinical maternal and fetal outcomes. Telemedicine also proved not be inferior to traditional methods in the management of GDM, reducing substantially the number of outpatient visits, but increasing at the same time the weekly checks of blood sugar levels using a virtual platform. We then proceeded to evaluate maternal satisfaction in the treatment of gestational diabetes, nding that women undergoing telemedicine follow up felt more controlled and that this method adapted better to their lifestyle versus those treated with the conventional method. It is the only study in the literature that studies maternal satisfaction by comparing the two methods using a speci c questionnaire on gestational diabetes, previously validated ad hoc.
The principal limit of our observational study is the small sample. Although the two groups were statistically homogeneous with each other, the increase in hospitalizations in NICU and in cases of respiratory distress for telemedicine group was generated by confounding factors in the cohort: the presence of two twin pregnancies and 17% of preterm births, which may have in uenced this outcome.
Furthermore, the questionnaire was administrated by telephone, and this may have inferred on the answers received, especially in the conventional group where not all women were native Italian speakers.
The non-inferiority result of telemedicine regarding the clinical maternal and fetal outcomes compared to conventional follow up is in line with ndings present in the literature. In fact, it is not clear which is the best method between telemedicine and standard care, according to a 2017 Cochrane review ( ve randomized controlled trials), where there are no clear differences in obstetric outcomes (preeclampsia or hypertension, caesarean section or induction of labor) or neonatal ones (fetuses LGA, hypoglycemia, severe morbidity and neonatal death) [14].
The increase in hospitalizations in NICU and in cases of respiratory distress is not supported by the literature, even if the recent meta-analysis by Xie et al of 2020 does not show a signi cant reduction in the risk of hospitalization in neonatal intensive care, jaundice or hyperbilirubinemia or NARDS neonatal acute respiratory distress syndrome in telemedicine follow up. The authors concluded that current evidence was not enough and further research comparing telemedicine to standard care is still needed to bolster the evidence [15].
To investigate maternal satisfaction, we used the OMDTSQ questionnaire created and validated on women treated with telemedicine by Hirst et al, 12 then we submitted it to both telemedicine and conventional groups for comparison. We understand that this is the only study in the literature that  Study selection, conducted in two phases. In the rst phase, the patients were enrolled among those diagnosed with gestational diabetes with positive OGTT (Oral Glucose Tolerance Test) divided into two groups: one controlled with telemedicine and one with the conventional method (outpatient visits) and were followed up until delivery (collection of obstetric, diabetic and neonatal clinical data). In the second phase, after more than six months from the birth, they were contacted by telephone by the same operator to answer a questionnaire on the satisfaction of the treatment of gestational diabetes (the modi ed Oxford Maternity Diabetes Treatment Satisfaction Questionnaire).