Psychological Care for Children and Adolescents with Diabetes and Patient Outcomes: Results from the International Pediatric Registry SWEET

,


Introduction
Te International Society for Pediatric and Adolescent Diabetes (ISPAD) Clinical Practice Consensus Guidelines (CPCG) notes that "being diagnosed with diabetes in childhood or adolescence can interfere with the normative developmental changes and interact with psychological and social factors in youth and their families. Integrated and collaborative care is, therefore, necessary" [1]. Terefore, the ISPAD guidelines recommend easy access to psychosocial care for children and adolescents with type 1 diabetes (T1D) and their families [1]. Similar recommendations are also proposed by other international and national diabetes societies [2,3]. Tese guidelines are based on the scientifc evidence which demonstrates that diabetes impacts nearly all aspects of an individual's life [4]. Furthermore, it is shown that blood glucose values are not only dependent on the administered amounts of insulin, however, they are also infuenced by numerous psychosocial factors [5][6][7].
Te efectiveness of psychological interventions for people with diabetes is established with primary research and meta-analyses reporting benefts resulting from providing individual patients with psychological care or with specifc interventions [8][9][10][11]. To our knowledge, there is little or no data showing whether the availability of psychological care in the individual diabetes center, is associated with the treatment outcomes and/or overall wellbeing of the center's patients.
Signifcant and sometimes substantial diferences between pediatric diabetes centers have been shown in the past [12,13]. Te Hvidoere Study Group confrmed that gender, age, and family support impacted on the individual variability of hemoglobin A1c (HbA1c) [14][15][16]. Tey then demonstrated a signifcant association between the HbA1c treatment targets reported by the staf and the centers' HbA1c results at the patient level [14][15][16]. Te latter result is replicated in two other studies [17,18]. Te Hvidoere Study Group also reported that another center-related factor associated with glycemic control was an efective collaboration of the multidisciplinary diabetes team members [15].
Although according to the ISPAD CPCG, children, adolescents, and young people with T1D and their families should have an easy and individualized access to psychological care, there is little published data to establish whether the availability of psychological services in pediatric diabetes centers is associated with patient outcomes. De Witt and colleagues [19] reported on a 2011 survey of ISPAD members in relation to their provision of psychological support. Tey reported that psychological care with an integrated mental health specialist occurred is less than half of the responding services, with larger centers more likely to provide integrated psychological services. Terefore, we decided to explore this question by using the data available from the multinational SWEET (Better control in Pediatric and Adolescent diabeteS: Working to crEate CEnTers of Reference) network and surveying them about their provision of psychological services [20]. Centers participating in SWEET, certifed as centers of reference and collaborative or associated centers are required to comply to with a clear criteria including following the ISPAD CPCG. Centers upload their data regularly, twice yearly, into a database. In addition to analyzing the availability of psychological care in SWEET centers, this study attempted to assess if access or diferent psychological service features/elements were associated with glycemic control.

Methods
All centers who participated in the SWEET database in 2020 (n � 112; data from the 2019 treatment year) were invited to complete a structured online 17-item survey (Google Forms, Google LLC, California, United States) regarding their psychology service. Te survey itself and the study fowchart are available online (Supplement and Supplementary Figure 1). Statistical analyzes included only centers with data in the SWEET database that provided answers to the questionnaire. Data were extracted from the SWEET database using the following criteria: 2019 treatment year and patients with T1D aged ≤18 years. Te following data were aggregated for each patient (median for continuous variables and maximum for binary variables): age, biological sex, age at diagnosis, body mass index standard deviation score calculated using the World Health Organization reference values (BMI SDS), daily insulin dose (U/kg), HbA1c (mmol/ mol) and (%), use of insulin pump (CSII), use of continuous glucose monitoring (CGM; includes both, real-time and intermittently scanned CGM), history of diabetes ketoacidosis (DKA; DKA at diabetes onset was not taken into account), and severe hypoglycemia (SH) as well as the number of self-measurements of blood glucose per day (SMBG). DKA and SH were presented as binary variables, and patients with 1 or more DKA or SH episode within the observation period was considered as patient with DKA or SH, respectively. Patients with missing data on pump use were excluded, while missing data on sensor use were considered as no sensor use. For all the other variables, missing data were only excluded from all analyses requiring the respective information. Questionnaire answers were grouped (detailed data available in the online Supplementary Data 1) and cross-tables were created.
Te centers with available/presence of any psychological care (referred to further as PsyC centers) ofered by a trained To measure associations with HbA1c (mmol/mol) and odds ratios (ORs) for DKA and SH related to survey responses, linear/logistic regression models were used. To take regional diferences into account, a random intercept for regions with an unstructured variance-covariance matrix and an optimization technique of Newton−Raphson with ridging was implemented. Regions were defned as follows: Asia and Middle East + Africa, North America, South America, and Australia + New Zealand. All regression models were implemented for aggregated data of each patient in 2019 and adjusted for age (categorized: <10 years, 10 to <14 years, and ≥14 years), gender, age at T1D onset (categorized: <6 years, 6 to <10 years, and ≥10 years), pump use (yes/no), number of SMBG (categorized: ≤4, >4, and CGM), center size (≤500 patients, >500 to ≤1000 patients, and >1000 patients), HbA1c target (categorized: ≤7%, >7% or ≤53 mmol/mol, and >53 mmol/mol), and completeness of documentation defned as data on ≥50% of patients available (yes/no). To adjust for multiple comparisons the Tukey−Kramer method was used.

Results
Responses were received from 76 (68%) centers in the SWEET database (uploaded data for a total of 27,819 subjects) with 52% male, 12.9 (IQR 9.7; 15.5) years old, age at T1D onset 7.3 (4.1; 10.5), BMI SDS 0.53 (−0.17; 1.25), mean HbA1c 62 (53; 74) mmol/mol or 7.8 (7.0; 8.9) %, and daily insulin requirement 0.8 (0.63; 0.98) U/kg. In the centers that responded to the survey, 45% of patients used CSII and 46% were current CGM users. In terms of acute complications, 2.4% of patients has experienced at least one DKA and 1.3% at least one SH episode in the analyzed period (1 year). Te characteristics of patients from these centers are comparable to that of all T1D pediatric patients from the SWEETregistry, and a detailed comparison is provided in the online Supplementary Table 1.

Availability and Characteristics of Psychological Care.
Psychological service was ofered in 89% of the centers that responded to our survey. Te characteristics of these centers and those without structured psychological services are shown in Table 1.
Te analysis revealed diferences in various aspects of psychological services depending on the center size. Tese are shown in Figure 1. Te lack of psychological support and consultations was more frequent in small than in mediumsized centers and was not reported by any of the big centers. A social worker as part of the MDT was present in all big and majority of small and medium centers. Te larger the PsyC center the more frequently the MHS was providing care exclusively for patients with diabetes. In large and medium centers, more often one MHS was responsible for more than 200 patients and families, whereas in small centers more fulltime equivalents (FTE) of MHS per 100 patients were observed. Psychological screening tools are used in a majority of big centers and approximately a quarter of small and medium centers.
In 75% of the PsyC centers, psychological services were fully covered and in the remaining quarter the patients needed to contribute fnancially, at least partially for the psychological consultation.
Te documentation of psychological consultations was integrated into the patients' medical records in 79% of PsyC centers. Te majority of PsyC centers ofered psychological care/support at T1D onset, mostly ofering more than one consultation (31%) or MHS contact adjusted to the patient's and the family's needs (39%). Only a few (6%) reported having psychological services available at T1D's frst appearance.
Practices regarding referral to a MHS difered between centers, and 40% of the centers psychological consultations took place at the request of the patient or their caregivers, and in 35% only on the request of the physician. In 25% of the clinics, the patients had at least one structured psychological consultation a year, with additional appointments scheduled at patients, physicians, or at other MDT member's request. Te latter approach was frequently reported by small and medium centers and predominantly in big centers ( Figure 1). Ongoing psychological services were available in most PsyC centers (78%), while the remainder ofered only a one-of contact with a MHS. One-third of the PsyC centers used the structured psychological questionnaire tools to screen for mental health issues.

Availability and Features of Psychological Care in Centers in Terms of Patient Outcomes.
Linear regression models showed that the availability of psychological services in centers was associated with a slightly lower HbA1c of the patients (68 vs. 72 mmol/mol, p � 0.004). Having no access to psychological care was associated with higher odds for DKA (OR with 95% CI: 1.8 (1.1-2.9), p � 0.027), but was not Subsequently, we analyzed the associations of diferent features of psychological care on the patient's outcomes: HbA1c, DKA frequency, SH rate, and BMI SDS. Signifcantly lower HbA1c values were observed for the following aspects of psychological care: MHS working only with patients with diabetes, documentation from psychological consultations added to the patient's medical record, available ongoing psychological care, available contact with an MHS at T1D onset, and no additional cost for psychological care (detailed results are shown in Table 2).
Regarding acute and severe diabetes complications we found that lower OR for DKA was signifcantly associated with no patient fnancial contribution to the costs of psychological consultations (0.8 (0.6-1.0), p � 0.049). No association of the rate of DKA was found with any of the following other variables describing availability and features of psychological care: who was the recipient of psychological consultations (exclusively patients with diabetes or also those with other conditions), the number of MHSs in the center, incorporating psychological consultations to the patients records, use of screening tools, type of referral for psychological consultations (any type compared to having a psychological consultation at least once per year with additional consultations upon referral by the patient or MDT members), presence or no psychological care at T1D diagnosis, and availability of ongoing psychological care.
Regression models revealed interesting associations of psychological care availability and its features with BMI SDS, as shown in online Supplementary Table 2 and Data 2. Tere were also diferences in sensor use depending on the structure of psychological services (online Supplementary Data 3).

ISPAD Recommendations regarding Psychological Care.
Only 4% of PsyC met all 4 recommendations related to access to a mental health specialist as a part of psychological care, as recommended by ISPAD CPCG. A further 15% met 3 recommendations and 34% met 2 recommendations. In 2 out of the 3 centers with full ISPAD CPCG compliance, the patients need to contribute fnancially at least partially to the psychological consultation. Te number of MHSs/100 patients as well as MHS FTE/100 patients did not difer between centers and were dependent on the number of care recommendations they met.
Regression models showed that adherence to ISPAD CPCG (categorized: available psychological care+ up to any 2 additional features recommended by ISPAD as described in methods and psychological care+ any 3 or all 4 of additional features, compared to no psychological care) was associated with lower HbA1c and BMI SDS (Table 3 and  online Supplementary Table 2 and Data 2). Increasing compliance with the guidelines was associated with decreasing OR for DKA and SH rates (Table 3).

HbA1c
Target. Te SWEET centers reported HbA1c targets ranging from 6.5% (48 mmol/mol) to 8.5% (69 mmol/mol), with <7% (53 mmol/mol) being the most common one. A lower HbA1c target (≤6.5% or 48 mmol/ mol) is more frequently applied in smaller centers, and big centers do not use it at all and more frequently set higher HbA1c targets. Some centers reported using age-related targets for their patients, despite standard ISPAD guidelines.
Although among all HbA1c target groups, there were PsyC centers as well as those with no psychological services available, those with the lowest HbA1c target (≤6.5%, 48 mmol/mol) universally ofered psychological care to their patients with MHS documentation to the medical record, and fully covered consultations costs (no patient contribution). However, none of the centers with target HbA1c ≤6.5% (48 mmol/mol) reported to comply with all ISPAD CGPG recommendations regarding the recommended features of psychological care.

Discussion
Te most signifcant fnding of this study was that access to psychological service in pediatric diabetes centers was associated with substantially lower rates of DKA. Whilst there was also a statistically signifcant association with HbA1c, this was not clinically meaningful. In addition, this study reafrms previous research demonstrating an association between pediatric centers treatment targets and HbA1c. Although this latter result is not novel, it extends the previous research of the Hvidoere Childhood Diabetes Study group, to a wider sample of diabetes centers.
Tis study is the frst to demonstrate an association between the availability of psychological care in a center and the outcomes of care for children and adolescents. Tere are clear mechanisms by which we would anticipate that having ready access to psychological care would result in reduced DKA. Te literature on recurrent DKA is consistent in identifying that this is commonly associated with missed or omitted insulin injections and concurrent mental health issues [21][22][23]. Early referral, easy access, and afordable psychological care would enable the diabetes service to engage early with individuals at high risk for DKA and or pick up mental health problems early. Tus, with an early initiation of psychological intervention for these individuals, it is likely to prevent individuals from progressing to the point of DKA. Combined with the discrete change in HbA1c Pediatric Diabetes it may suggest that early psychological intervention allows to prevent the deterioration of glycemic control. MHSs, as members of the MDT may also help in retaining the motivation of patients. Tey may also help other MDT members to better adapt the treatment to the needs of specifc patients and individualization of diabetes care was already described to be positively associated with glycemic control [15]. Yet the exact mechanisms need to be investigated further and these were not the aim of this study.
Te data reported here provide an important initial evidence supporting the current recommendations regarding the availability of psychological services. Whilst there are many studies demonstrating the efectiveness of specifc psychosocial interventions [24,25], these trials rarely have samples representative of the general clinic population [26][27][28][29]. Tis study reports on the impact of the availability of psychological care for children with T1D in clinical practice. Tus, these data can be used to extrapolate the benefts of integrating psychological services into the pediatric diabetes care, and advocating for coverage of costs to remove the potential fnancial burden on families.
Te fndings of lower HbA1c targets are associated with a better glycemic control of the patients [14,15,17,18]. Te analysis of treatment targets and psychological care availability leads to a second point that is critical to note: having access to psychological care and meeting the ISPAD guidelines does not mean taking "a soft approach" and pursuing less demanding treatment targets for children and adolescents with diabetes. Centers with easy access to psychological care do not have less demanding treatment targets and these centers have individuals with lower HbA1c levels. Tus, psychological care is arguably enabling people to succeed, and it is supporting wellbeing as well as glycemic outcomes. Lastly, this study replicates the fnding that centers that strive to achieve lower HbA1c targets are achieving better outcomes with their patient population. Clearly, there is a need to understand why some centers do not aim for lower HbA1c targets, and how do we support the adoption of current standards of care into widespread clinical practice.
Te features of psychological care that are related to a better access of the patient to a MHS, such as available ongoing care, more than one MHS working in the center, as well as sharing the conclusions from psychological consultations with the whole MDT (adding them to the patients' documentation), were associated with better outcomes. It may suggest that psychological interventions when available may be introduced earlier, that is, before the patient's metabolic control deteriorates. It can be also hypothesized that due to the information from the MHS, other MDT members have a more holistic picture of the patient's life with diabetes and therefore adapt the training and communication to the diferent degrees of motivation and acceptance of diabetes. Further studies would be needed to provide evidence for these concepts.  Te majority of SWEET centers ofer some psychological care to their patients, with varying degrees of implementation of all recommendations from ISPAD CPCG regarding psychological care. It is not possible to compare these results with the previous survey of ISPAD members, as the guidelines and questions have changed and the sampling frame is substantially diferent. However, they do not suggest that there has been a substantial uptake of psychological services in pediatric diabetes centers, with only 4% of centers meeting the 4 guideline statements about access to a mental health specialist. Tere were signifcant diferences in the organization of these services between centers, and one of the factors was the center size; and larger centers ofer more psychological care. Tis replicates the previous survey of the ISPAD membership showing that larger centers were more likely to have integrated psychological services [19]. If psychological support was organized more consistently the OR for DKA and SH would be lower, but the diference in HbA1c and BMI would not be clinically relevant.
Tis is the frst study to assess the overall association of psychological care availability on metabolic outcomes of pediatric patients with T1D. Te results are based on the analysis of a large worldwide database from centers that participate in a twice-annual benchmarking and share the same goals for diabetes care. In addition, the response rate to the questionnaire among SWEET centers was high and therefore the studied group of centers can be considered representative. A limitation of this study is the discrepancy in the number of patients from centers with and without available psychological care, but regression models were adjusted for center size. Further limitations that need to be acknowledged include the following: lack of information on the percentage of patients receiving psychological care in single centers and its nature (ongoing care, support, and single consultation), and potential other factors related to the MDT composition. We acknowledge that other factors, for example, related to the approach to the patient or means of provided care could infuence patient outcomes. Although the results were adjusted for multiple confounding factors, one cannot exclude that centers which implement psychosocial guidelines, have a diferent approach to diabetes care. Future studies should be designed to investigate the mechanism by which the psychological services impact patient outcomes.

Conclusions
In summary, most centers from the SWEET registry ofer psychological care consistent with CPCG recommending easy access to psychosocial care for children and adolescents with T1D and their families. Te data reported here suggests that having psychologists as part of the multidisciplinary team is associated with lower rates of life-threatening and compromising events such as DKA, and may be a funding priority for centers without access to such services.

Data Availability
Te data that support the fndings of the study are available from the corresponding author upon request.

Ethical Approval
SWEET (NCT04427189) was approved by the Ethical Committee of Hannover Medical School. All contributing centers fulfll current regulatory data protection security (pseudonymized data collection) and ethics compliance.

Disclosure
Parts of this study were presented as abstracts during the 14th International Conference on Advanced Technologies & Treatments for Diabetes 2-5.06.2021 (virtual meeting) and the 47th International Society for Pediatric and Adolescent Diabetes Annual Conference 13-15.10.2021 (virtual meeting) [30].
especially the collaboration centers in this investigation (Appendix). SWEET is a registered nonproft charity in Hannover, Germany. It is fnanced through membership fees of the participating centers (based on income of country of residence according to the World Bank) and corporate members. Te authors acknowledge with gratitude the support from the following SWEET e.V. corporate members in an alphabetical order: Abbott, Boehringer Ingelheim, DexCom Inc., Insulet, Eli Lilly & Co., Medtronic Europe, and Sanof. Te Sweet Project is an ongoing registered research collaboration (NCT04427189).

Supplementary Materials
Supplementary Figure 1: fowchart for selection of the study population from the SWEETregistry. Supplementary Data 1: grouping of the questionnaire answers. Supplementary Table 1: characteristics of patients with type 1 diabetes aged <18 years from all SWEET centers in the 2020 database and patients from canters that responded to the survey. Supplementary Table 2