Pre‐empting the challenges faced in adolescence: A systematic literature review of effects of psychosocial interventions for preteens with type 1 diabetes

Abstract Introduction Numerous psychosocial interventions have been conducted in children and adolescents with type 1 diabetes, aiming to improve their self‐management and autonomy acquisition. However, these tend to address family conflict and parental perspectives, and a scarce number of interventions explore the outcomes among preteens. This review examined the outcomes of psychosocial interventions for preteens with type 1 diabetes, as an under‐researched field to date. Methods A systematic literature review of intervention studies with randomized controlled trial design, targeting preteens with type 1 diabetes, was conducted. Six databases were searched for publication periods from 1995 to October 2019. Quality of the interventions according to the International Society for Pediatric and Adolescent Diabetes (ISPAD), as well as reporting and effect sizes, were assessed. Results Twelve studies were selected, covering ten interventions. According to the topics identified, four of these interventions were categorized as self‐care programmes, three as psychosocial programmes and three as mixed. All of the interventions, except for one, covered ≥50% of the ISPAD recommendations. Reporting adequacy was negative only in one intervention. Main outcomes were glycemic control and self‐management, but effect sizes could only be calculated for half of the interventions with no overall significant effect. Conclusions This review shows a lack of adequate psychosocial interventions targeting preteens with type 1 diabetes and actively involving them as participants. These intervention's educational programmes and methods should be standardized to guarantee successful results. New technologies and peer support implementation could be a promising pathway when designing these studies.


| INTRODUC TI ON
Prevalence of Type 1 Diabetes (T1D) has alarmingly increased worldwide during recent years, 1 with the highest rates registered among children in Finland and Sweden (57.6 and 43.1 per 100 000 per year for children aged under 15 years old, respectively). 2 Left untreated, persistently elevated blood glucose levels are associated with long-term complications such as clinical retinopathy, nephropathy, neuropathy and vascular disease. 3 Achieving optimal diabetes outcomes may become challenging during different stages of life. Adolescence represents a critical period for children with T1D, in that it is associated with an increased likelihood of experiencing higher blood glucose levels and recurrent hypoglycaemia and ketoacidosis episodes, which burden adolescents' burgeoning autonomy and diabetes self-management. 4,5 Numerous interventions have been conducted aiming to address these problems with limited success in terms of clinical and psychosocial outcomes. [6][7][8] This inefficacy has been suggested to be related to the timing of these interventions, usually during adolescence, 9,10 and their tendency to target family conflict and parental perspectives. Some challenges around engaging adolescents during an intervention involve the limitations when taking the theory to practice, the difficulties with adherence and follow-up and the use of the right methods tailored to this life period. 11 Streisand and Mednick 10 discuss the potential positive effects psychosocial interventions can have on self-management behaviours and metabolic control during adolescence when interventions are administered starting in the preteenage period.
Family interactions play an essential role in diabetes management among preteens and may influence self-care and glycemic control during this developmental period. [12][13][14] Likewise, a broad range of studies have reported changes in parental stress and quality of life when a child is diagnosed with diabetes. [13][14][15] However, less research can be found evaluating this impact among preteens. 16 Preteens are a term that includes children within the life-span of 9-12 years old. 17 For this reason, the present systematic review examines the outcomes of psychosocial interventions for preteens with T1D, the aim being to inform the intervention research on preteens with type 1 diabetes. A systematic literature review was conducted followed by an analysis of the selected studies' results.

| ME THODS
To assess the results of interventions conducted among preteens with T1D, a systematic literature review was conducted. This review protocol was not preregistered. Psychosocial interventions included teaching diabetes-related knowledge or skill, psychosocial training or support as well as psychotherapeutic interventions targeting individuals and families. 7 Psychosocial aspects covered behavioural, psychological and social issues in relation to living with diabetes, for example self-management, coping and communication.

| Search strategy
A literature search was conducted in six databases: PubMed, PsycINFO, CINAHL, Web of Science, SCOPUS and Sociological Abstracts, applying a criterion for publication periods from 1995 to October 2019. This time range was based on previous systematic reviews in the field. 9,18 Two searches were conducted for each database using Medical Subject Headings (MeSH) terms and Boolean operators, being 'Diabetes Mellitus, Type 1' AND 'Child' AND ('Self Care' OR 'Self-Management') for the first search, and 'Diabetes Mellitus, Type 1' AND 'Child' AND ('Psychology' OR 'Sociology' OR 'Anthropology') for the second search.
These are very broad terms used to ensure that the search captured the right age span as well as an extensive range of psychosocial issues, for example mental health and quality of life.
All of the articles found were imported to EndNote and classified by database. The results obtained were compared among databases in order to discard duplicates.

| Inclusion and exclusion criteria
No restrictions on article language were applied. The following inclusion criteria were applied: (a) type 1 diabetes, (b) age span of 7-13, (c) involvement of preteens as intervention participants, (d) any kind of psychosocial aspect, (e) published from 1995 to 2019, (f) intervention study and (g) randomized-control trial (RCT). Studies not involving preteen's participation and/or based on types of diabetes other than type 1 as well as other conditions were excluded. Articles that did not report any research data or only reported nonpsychosocial outcomes for the review were excluded (such as protocols or feasibility and acceptability studies).

| Screening process
The screening process consisted of three stages: (a) a first screening for title and abstract was carried out by one coder applying criteria 1-5. If it was unclear from the abstract whether papers met the inclusion criteria, full paper manuscripts were obtained; (b) a second screening was conducted by two coders applying criteria 1-7, and new articles were identified by reviewing reference lists. When the studies did not meet the criteria or when this was unclear, a com-

TA B L E 1 (Continued)
intervention details of the selected articles can be found in Tables 1 and 2, respectively.

| Risk of bias assessment
Each of the interventions was assessed for risk of bias according to the Cochrane Collaboration's tool. 19 This assessment consists of the following seven criteria: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and researchers (performance bias), blinding of outcome assessment (detection bias), incomplete outcome assessment (attrition bias), selective reporting (reporting bias) and other bias due to problems not covered within the previous criteria. Each of these criteria is judged as low risk of bias, if the bias is unlikely to alter the results seriously, unclear risk of bias if this bias raises some doubt about the results and high risk of bias if the bias may alter the results seriously (Table 3).

| Primary quality assessment
The quality of the educational interventions was assessed using Every intervention was assigned a global score out of 19 based on the number of recommendations met, followed by a second score by category identifying the weak points (Table 4). A second score <50% indicates the need to improve the educational programme in that category in relation to the ISPAD guidelines. 20 The quality assessment was conducted by one coder and checked by three other coders.

| Reporting assessment
Reporting was assessed based on the four elements checklist for quality assessment elaborated by Carroll et al 24 as this review started out including both qualitative and quantitative studies and was then reduced to only RCTs. Each publication was reviewed to determine whether the question and study design, participants' recruitment and selection, and the methods of data collection and analysis were reported adequately. Subsequently, the studies were dichotomized as adequately reported if they received a Yes on two or more criteria, and as inadequately reported if they were assigned a Yes on one or fewer criteria.

| Effect sizes
Effect sizes were calculated as the standardized mean difference (SMD) between control and intervention groups for selected outcomes. These, together with the P values, are shown only for the interventions which provided enough information to allow for these calculations ( Table 5). As suggested by GRADE, a cut-off point of 0.5 for SMD was used as a rule of thumb for an important effect size difference. 25

| RE SULTS
The process leading to the selection of the 13 studies for the review, covering 11 interventions, can be found in Figure 1. 26 Two articles covered results from the same intervention at different times on two occasions. [27][28][29][30] One of the interventions was a clusterrandomized trial. 31 The remaining 10 were RCTs, of which only six provided information about randomization, and coin toss was the method used in one of them. 32 In addition, the paper by Sullivan-Bolyai et al 17 was a feasibility study; the authors were contacted to obtain additional information about their feasibility study and confirmed that no further data on this intervention were published after this paper.  Table 1). The interventions were primarily guided by social cognitive theory and motivational interviewing.
These 11 interventions covered a total population of 1659 children between 4 and 17 years of age, always including the 7-13 preteen age span criteria. Two of the studies analysed the results in two age groups, being 9-11 and 12-14 years old 33 and 8-12 and 13-16, respectively. 34 All participants had type 1 diabetes diagnosed at least 3 months, 33

| Delivery of the intervention
Intervention details can be found in Table 2. Seven of the interventions took place in a healthcare setting (hospital, clinic), and 1 was exclusively home-based. 32 Three trials gave participants the option to choose among undergoing the intervention at home, in a community setting or at a public location. Delivery modes were individual (n = 7), group (n = 2) or combined (n = 2), always including preteens as participants. Parents were included (n = 5) or free to attend (n = 6).
Four of the interventions were carried out by a multidisci- conducted two of the interventions, one of which was supported by nonprofessionals belonging to the preteen's environment (ie parents, teachers). The remaining interventions were delivered by a group of trained nonprofessionals (students) (n = 2), an adolescent mentor together with a nurse advisor (n = 1), a health advisor (n = 1) and a robot (n = 1).
Interventions lasted from 1 day (n = 1) to 18 weeks (n = 1), and 2 (n = 2), 12 (n = 4) and 24 (n = 1) months. One of them was currently ongoing on the publication date, and another one was unclear in

| Glycemic control
Glycemic control was evaluated by measuring blood glucose in one of the interventions, 36 and glycosylated haemoglobin (HbA1c) was the primary outcome in over half of the interventions (6 out of 11).

| Self-efficacy
Self-efficacy was measured using the Self-Efficacy for Diabetes Scale in Confidence in Diabetes Self-Care (CIDS) scale to measure diabetes management self-efficacy beliefs. This is a 20-item self-report questionnaire, addressed to patients with type 1 diabetes, assessing self-efficacy as the perceived ability to perform diabetes self-care tasks. 41 Only the intervention by Grey et al showed benefits on self-efficacy.

| Coping skills
Coping was measured in one intervention 29

using the Issues in
Coping with T1D-Child Scale. This scale consists of 12 items inside of 2 subscales rated on a 4-point Likert-type scale, the aim being to assess perceptions of how hard or difficult to handle and how upsetting T1D management is. The intervention showed no differences between groups regarding this outcome.

| Quality of life
The

| Psychological distress
The

| Intervention acceptance and/or satisfaction
Acceptability of the intervention was measured using study-specific nonvalidated surveys rating the programme's participants' satisfaction

TA B L E 4 Matching with ISPAD Guidelines
Type of programme

Fiallo-Scharer et al (2019)
It is run at a location accessible to individuals and families, whether in an ambulatory setting or not It uses a variety of teaching techniques, adapted to meet the different needs, personal choices, and learning styles of youths with diabetes and their parents authors were contacted for further information; they confirmed that the study data were no longer available and that the final outcomes were never published. 10

| Risk of bias assessment
Assessment of risk of bias can be found in Table 3. This was a complicated process due to the lack of randomization details in 5 of the 11 interventions, that difficulted the assessment of selection bias as random sequence generation and allocation concealment criteria.
Moreover, blinding was not specified or partly described in many occasions, leaving the risk of bias as an unclear statement in 6 out of the 11 interventions. The highest risk of bias (in 5 out of 11 interventions) was indeed found for the detection bias as blinding of outcome assessment, as researchers assessing the outcomes were frequently nonblinded. The criterion where the lowest risk of bias was found was the one regarding reporting bias as selective reporting, due to most of the interventions adequately reporting all the prespecified outcomes (7 out of 11). Similarly, criteria regarding both attrition bias as incomplete outcome data and other bias accounted for a low risk of bias in 6 out of 11 interventions. Drop-out rates were low in general and some of the interventions described a method to deal with this missing data that is intention to treat analysis, and most of the bias was covered by using the Cochrane Collaboration's tool. was the one carried out by Fiallo-Scharer et al 34 All of the interventions covered >50% of the recommendations except one of them, which met 9 out of the 19 criteria. 34 An analysis of the three categories stated by the ISPAD Guidelines was conducted to facilitate the identification of the strengths and weaknesses of every intervention.

| General recommendations
The main points that all the programme types lacked were the use of new technologies (2 out of 10) and the presence of an interdisciplinary team delivering the intervention (4 out of 10). The remaining recommendations were present in most of the programmes.

| Universal principles
The main reason for downgrading a point on this category was the recommendation regarding the right to comprehensive expert, structured education for every young person, owing to the diabetes duration criterion and other exclusion criteria such as the most recent HbA1c levels [36][37][38] and being a girl with menarche. 17 All of the programmes accomplished at least 50% of the universal principle's recommendations.
F I G U R E 1 Flow chart of the screening process

| Characteristics of a structured programme
The fewer recommendation matches found for this category were for the audition (3 out of 10) and the peer groups or school friendship presence (3 out of 10), followed by the quality assurance (5 out of 10) of the programme delivered and the variety of teaching techniques used in the intervention (5 out of 10).

| Reporting assessment
Of the 11 interventions, all qualified as being adequately reported except for the paper by Streisand and Mednick, 10

| Effect size
A summary of effect sizes can be found in Table 5 Regarding the remaining interventions, no numerical data were available for calculating SMD, or the data were incomplete, as in Nansel et al 2007Nansel et al , 2009 where SD was missing for HbA1c outcomes. Similarly, the intervention by Lasecki et al 36 showed effect sizes for every participant without comparison between groups, and no SD was provided, meaning that these calculations could not be confirmed.

| D ISCUSS I ON
Results from this systematic literature review indicate that there have been only a few trials of psychosocial interventions targeting and actively involving preteens in the age range 9-12 years when compared with the number of trials for adolescents/teenagers.

Comparison among interventions became complicated owing
to the wide and heterogeneous variety of educational programmes, frequently with nonstandardized methods and measurement instruments, as it could be seen when assessing the risk of bias ( Table 3).  (13-for increasing diabetes knowledge, but corresponded to feasibility results for an intervention that, to our knowledge, was never published. Thus, besides having a well-structured programme, these findings suggest the need for educational programmes tailored to the age of the target population, especially when this may cover two life stages (pre-adolescence and adolescence) and young people who can be divided into two groups that might benefit from receiving a specially adapted intervention. As already stated, the oppo- was not considered in the majority of interventions, as a minimum of 6 months duration of diabetes was established as a criterion for 90% of the interventions, being 4-6 years duration of diabetes in one occasion. 36 The argument for this criterion application was explained in Pendley et al, referring to the 'honeymoon' period after diagnosis, where small doses of exogenous insulin are needed as the pancreatic cells are still able to produce insulin, and thereby blood sugar changes cannot be exclusively associated with self-management behaviours. However, diagnosis usually occurs during childhood and pre-adolescence, so recently diagnosed children might benefit from psychosocial interventions that prepare them for the upcoming diabetes management and improve self-efficacy outcomes in the future. 47 Some studies have even revealed the possibility of a slowdown, aiming to stop the remission when combined with pharmacological therapies. 48,49 Other criteria, identified as interfering with the above-mentioned universal principle, were the presence of menarche as an exclusion criterion 17 and the most recent HbA1c% reported, the main aim being to recruit participants with poor glycemic control. Nonetheless, conducting this type of intervention in well-regulated preteens with type 1 diabetes can take advantage of peer support, as already stated, because these preteens can learn new techniques and skills to be transmitted, through a role-model function, to peers with diabetes who have lower metabolic control, as pointed out by Colson 20 regarding Bandura's social learning theory. 50 Furthermore, the country of intervention can play an important role when carrying out an educational programme. Only 2 of the 11 interventions assessed were conducted in a country other than the United States, making it difficult to generalize these results to preteens worldwide. This highlights the importance of studying preteen's needs and points of view, but also their ethnicity and race if we are to adapt the intervention to the implementation context. Likewise, socioeconomic status affects diabetes morbidity 51 and is highly connected to the use of healthcare services, 52 though it was only measured by one of the interventions with no significant results. 31 Grey et al also mentioned this factor, suggesting an association between their positive results on HbA1c and the medium-to-high socioeconomic status of the participants.
Besides this educational programme classification and the ISPAD Guidelines, comparison problems remained when assessing the methodology used. A heterogeneous variety of questionnaires and scales were used to evaluate diverse intervention outcomes, which were common among interventions on a few occasions, but usually modified to address the target population. 10,33 The need for a standardized methodology, specifying clear measurement instruments and outcomes to assess, becomes evident when examining the results of the present review.
Along with these findings, the limited number of psychosocial interventions conducted to date among preteens with type 1 diabetes it is a matter of concern, especially those actively involving preteens as participants, one of the main exclusion criteria used in the present literature search. Likewise, preteen's needs, feelings and points of view should be considered when designing these interventions, while including family and peers. This combination would allow preteens to achieve meaningful behavioural change, along with emerging diabetes management autonomy, that will form a foundation on their way to a good metabolic control and psychosocial well-being during adolescence and, consequently, adult life.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
Rey Velasco, E. undertook the data collection, data analysis and wrote the first draft of the manuscript. Pals, RAS acted as a second reviewer for study selection and supported data analysis and the final draft of the manuscript. Skinner, T. provided crucial academic input on the search and interpretation of the data and supported data analysis and the contents of the manuscript. Grabowski, D. conceived the review, provided academic supervision, and supported data interpretation, data analysis and the contents of the manuscript.

E TH I C A L A PPROVA L
Since this is a systematic review, ethical approval was not required.