The impact of hypoglycemia on quality of life and related outcomes in children and adolescents with type 1 diabetes: A systematic review

Objective To conduct a systematic review to examine associations between hypoglycemia and quality of life (QoL) in children and adolescents with type 1 diabetes. Methods Four databases (Medline, Cochrane Library, CINAHL, PsycINFO) were searched systematically in November 2019 and searches were updated in September 2021. Studies were eligible if they included children and/or adolescents with type 1 diabetes, reported on the association between hypoglycemia and QoL (or related outcomes), had a quantitative design, and were published in a peer-reviewed journal after 2000. A protocol was registered the International Prospective Register of Systematic Reviews (PROSPERO; CRD42020154023). Studies were evaluated using the Joanna Briggs Institute’s critical appraisal tool. A narrative synthesis was conducted by outcome and hypoglycemia severity. Results In total, 27 studies met inclusion criteria. No hypoglycemia-specific measures of QoL were identified. Evidence for an association between SH and (domains) of generic and diabetes-specific QoL was too limited to draw conclusions, due to heterogenous definitions and operationalizations of hypoglycemia and outcomes across studies. SH was associated with greater worry about hypoglycemia, but was not clearly associated with diabetes distress, depression, anxiety, disordered eating or posttraumatic stress disorder. Although limited, some evidence suggests that more recent, more frequent, or more severe episodes of hypoglycemia may be associated with adverse outcomes and that the context in which hypoglycemia takes places might be important in relation to its impact. Conclusions There is insufficient evidence regarding the impact of hypoglycemia on QoL in children and adolescents with type 1 diabetes at this stage. There is a need for further research to examine this relationship, ideally using hypoglycemia-specific QoL measures.

Introduction Type 1 diabetes is one of the most common chronic conditions among children and adolescents and requires a demanding treatment regimen (e.g., insulin administration several times a day, monitoring of glucose levels and regulation of food intake and physical activity) [1,2]. The goal of diabetes management is to achieve and maintain recommended glycemic levels to prevent/delay acute and long-term complications [1]. However, treatment with insulin can lead to hypoglycemia (low blood glucose level) [3]. Hypoglycemia can cause immediate uncomfortable symptoms (e.g., shakiness, dizziness), and in severe cases lead to confusion, seizures and coma, where self-treatment is not possible. In addition, recurrent episodes of severe hypoglycemia (SH) have been associated with neurocognitive impairments, especially in young children [4].
Although rates of SH in children and adolescents have decreased significantly in the past two decades, due to improvements in insulin administration and monitoring technologies (e.g., continuous subcutaneous insulin infusion and continuous glucose monitoring) [5][6][7], a recent systematic review still reported an incidence of 1.21-30 events per 100 person-years in young people with type 1 diabetes [8]. Hypoglycemia is particularly challenging and complex to manage in children and adolescents with type 1 diabetes for several reasons: this group has less predictable eating, activity, and sleep patterns relative to adults; children's diabetes is often (co-)managed by the parent; and young children may be unable to communicate their symptoms and needs [9]. Among adolescents, both hormonal changes leading to insulin resistance [10] and developmental changes, such as seeking independence from parents, that add to the burden of self-management, can lead to greater fluctuations in glucose levels and increase the risk of hypoglycemia [11].
Another important goal of pediatric diabetes management is to achieve and maintain optimal quality of life (QoL) [12]. While some studies have shown that hypoglycemia is negatively associated with QoL [13,14], other studies have not found such an association [15,16]. Although QoL is defined and assessed in many different ways across studies, it is recognized that QoL is a multidimensional, dynamic and subjective construct [17]. It has been argued that, to understand the impact of a condition on QoL, we need to ask people how satisfied they are with the areas of life that are important to them for their overall QoL, and then ask how these areas are affected by the condition, such as diabetes or, more specifically, hypoglycemia [18,19]. It is therefore important to critically examine the range of patient-reported outcomes (PROs) used in studies, and to determine which are measuring the impact on QoL, and which are measuring related outcomes (such as diabetes-specific emotional distress or health status) rather than QoL [18]. Synthesis of the current evidence base is needed to determine the relationship between hypoglycemia and QoL-related outcomes.
Therefore, our aim was to conduct a systematic review to summarize and critically appraise the evidence regarding the association between hypoglycemia and QoL (and related outcomes) in children and adolescents with type 1 diabetes.

Search strategy
This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20] and was registered on the International Prospective Register of Systematic Reviews (PROSPERO; CRD42020154023) database. A systematic search of Medline, Cochrane Library, CINAHL and PsycINFO databases was conducted in November 2019 and updated in September 2021, as part of a larger search strategy for five related systematic reviews examining the impact of hypoglycemia in various populations. Search terms included free-text and subject heading terms relating to the following concepts, separated by the Boolean operator "and": (1) type 1 diabetes, (2) children and adolescents, (3) hypoglycemia and (4) QoL and related outcomes. There were no limits applied to date or language at the search stage. The search string is provided in S2 File.

Inclusion & exclusion criteria
Studies were eligible if they: (1) included children and/or adolescents with type 1 diabetes, majority aged �18 years (or mean age <18 years old), (2) assessed the history of hypoglycemia, (3) included outcomes of generic, diabetes-specific or hypoglycemia-specific QoL (or domains of QoL) or related outcomes (e.g., fear of hypoglycemia, depression, diabetes distress), (4) examined the association between hypoglycemia and QoL or related outcomes, (5) had a quantitative design, (6) were published in a peer-reviewed journal with full text available in English, (7) were published after 2000. The focus was limited to publications in the past two decades, as diabetes management strategies and rates of hypoglycemia have changed considerably in recent decades [5][6][7]. Studies were excluded if they: (1) focused on cognitive functioning [21] or neurodevelopmental disorders [22], or (2) only included proxy-report (e.g., by parents) of outcomes [23].

Screening, article selection, and data extraction
Abstract screening was completed by three reviewers, with 10% of the abstracts being double screened (AS, AC and MCL). MC completed full text-screening (with input from a second reviewer (MB) where queries arose), and 10% of the full-text records were independently screened by a third reviewer (KM). In case of disagreement, reviewers discussed until consensus was reached. Additionally, forward chaining (i.e., citation searching of included studies in Google Scholar) and backward chaining (i.e., reference list checking of all included studies) was undertaken to identify additional eligible papers. Data extraction was performed by MC and KS; extracted data included reference details, study details, participant characteristics, analysis, results and discussion points. Extracted data were checked by two independent reviewers (HC, MB) and consensus was reached in case of discrepancies.

Risk of bias assessment
Risk of bias was assessed (MC) by the analytical cross-sectional studies critical appraisal tool from the Joanna Briggs Institute (JBI) [24]. Risk of bias assessment was not used to exclude studies but was discussed and summarized to aid interpretation of the quality of the evidence base.

Data synthesis
Narrative synthesis was structured primarily by a conceptual framework of QoL (Table 1), wherein outcomes were grouped based on two dimensions [18]: 1) the scope of the measure (global, broad, specific): i.e. whether a questionnaire assesses global QoL (e.g., 'overall QoL'), a broad domain of QoL (e.g., social functioning) or a specific domain of QoL (e.g., friends); and 2) the attribution of the measure (generic, diabetes-or hypoglycemia-specific): generic QoL measures ask people to rate areas of their life overall. These ratings can be affected by many factors including but also unrelated to diabetes, whereas diabetes-specific [or hypoglycemiaspecific] QoL measures seek to attribute any impact to the condition, specifically asking: 'how does diabetes [or hypoglycemia] impact on your QoL?' If a measure did not assess QoL but assessed a concept closely related to QoL (such as depressive symptoms, diabetes distress, FoH), it was classified as a related outcome. Diabetes distress refers to the negative emotions related to living with diabetes [25]. These related outcomes were grouped by the attribution of the measure (generic, diabetes-or hypoglycemiaspecific outcome). Within each outcome group, study findings were summarized separately for SH and non-severe hypoglycemia (NSH) according to the authors' definitions, and where possible by outcome type/questionnaire.
To enhance consistency and transparency in the narrative synthesis and to avoid vote counting based on statistical significance [26], the following considerations were taken into account to interpret the evidence per outcome: 1) whether definitions and recall periods of hypoglycemia varied between studies; 2) whether there was a valid assessment of QoL and/or related outcomes; 3) whether analyses were conducted for children and adolescents separately or together; 4) whether there were exclusion criteria that directly related to hypoglycemia or QoL outcomes; and 5) whether effect sizes were available: small (r�0.10 and <0.30), moderate (r�0. 30 and <0.50), and large r�0.50) [27].
It was determined that there was insufficient evidence to draw a conclusion for an outcome if: a) there were less than three studies examining the association, or b) there was considerable heterogeneity in definitions of hypoglycemia and sample characteristics across studies.

Included studies
The searches yielded 1165 results. Title and abstract screening resulted in 217 potential includes. After full-text screening, 17 studies were included. An overview of the full-text papers that have been assessed for eligibility with reasons for exclusion is provided in S1 Table. Forward and backward chaining yielded 10 extra includes. In total, 27 studies were included for data extraction and synthesis. Fig 1 provides an overview of the screening and selection process.

Study characteristics
The 27 studies included a total of N = 141,530 participants, with sample sizes ranging from N = 39 to N = 2,602, with the exception of two large-scale studies (N = 53,986 and N = 75,258). The studies were conducted in 18 countries, the majority conducted in USA (n = 6) and Germany (n = 4). One study included multiple countries in Europe, North America and Japan [37]. The age of participants ranged from 5-25 years. Most studies included participants between 8-18 years, although eight studies also included participants above 18 years, and four studies also included participants aged 5-7 years. One study included children aged 6-12 years and five studies included adolescents aged 12-18 years. Study characteristics are detailed in Table 2.

PLOS ONE
All studies had a cross-sectional design. Assessment of hypoglycemia relied mostly on retrospective self-report (n = 19, 70%), with recall periods ranging from the past month (n = 4) to the Table 1. Overview of quality of life and related outcome measures in the included studies, by breadth and attribution.

Quality of Life (QoL) and related outcomes Generic
Diabetes-specific Hypoglycaemia-specific (no attribution) (attribution to diabetes) (attribution to hypoglycaemia)

PLOS ONE
past three (n = 5), six (n = 5) or 12 months (n = 7), to period since diagnosis (n = 1). Hypoglycemia was reported by the child or adolescent with diabetes (n = 10, 34%), their parent(s) (n = 7, 24%), or both (n = 1, 4%); or based on a combination of self-report and glucose meter data (n = 2, 8%). Six studies (22%) focused on hypoglycemia that involved coma or seizures, assessed via medical records. Two studies (7%) did not specify how hypoglycemia data were derived. Twenty-one studies (78%) examined SH, defined as episodes: 1) where assistance of others was needed (n = 4); 2) resulting in confusion or seizures/coma (n = 7), or 3) characterized by a combination of these definitions (n = 8). Four studies did not specify a definition of SH. Ten studies (37%) examined "moderate" hypoglycemia, which will be referred to as "non-severe" hypoglycemia (NSH) throughout this review. Although it is recognized that NSH is usually referred to as "self-treated" hypoglycemia [51], that is not appropriate when describing hypoglycemia in (younger) children, as their parents often need to help regardless of the severity of the episode. Twelve studies (44%) used a continuous measurement of hypoglycemia (e.g., frequency of SH in the past 6 months), and fifteen (55%) reported on categorical measurement of hypoglycemia (e.g., absence or presence of SH).
The studies included 19 instruments to assess QoL and related outcomes ( Table 1). The most commonly used were the Hypoglycemia Fear Survey Child Version (HFS-C) (n = 6) and the Pediatric Quality of Life Inventory (PedsQL) Generic (n = 4) and Diabetes (n = 6) modules. An overview of all scales being used in the studies can be found in S2 Table.

Risk of bias assessment
Of the 27 studies, 93% included participants with a medically-verified diagnosis of type 1 diabetes. Most provided adequate details of their inclusion and exclusion criteria (89%) and

PLOS ONE
participants and settings (74%). In 52% of the studies, hypoglycemia was defined in accordance with the current International Society for Pediatric and Adolescent Diabetes (ISPAD) definition; namely, SH as an event with severe cognitive impairment (including coma and seizures) requiring assistance by others, and NSH as events with a blood glucose value �3.9 mmol/L (70 mg/dL) [52]. Most studies (81%) used statistical analyses appropriate to their data and, while all studies identified confounding factors, 67% adjusted analyses accordingly. Most studies (70%) used psychosocial outcome measures that were psychometrically valid and reliable instruments for use with children and adolescents with type 1 diabetes. In 11% of the studies, both validated and non-validated measures [13,28,36] were used, while 19% included measures that were not validated in adolescents with type 1 diabetes [14,34,35,43,47]. A full overview of the risk of bias assessment is presented in S3 Table. Narrative synthesis Table 3 provides a summary of the main findings of each study.
Global QoL. Table 1 shows that both generic measures and diabetes-specific measures of global QoL were used. One study used a non-validated questionnaire [13]. The age of the participants in these studies ranged from 11-21 years, and one study conducted analysis for children and adolescents separately [30].
Severe hypoglycemia. Three studies examined the relationship between SH and generic QoL using the KINDL-R [13] or the PedsQL [30,31]. One study showed that those with SH (not further defined) in the past month reported significantly lower QoL than those without SH, but this was not observed for SH in the past year [13]. Two studies found no significant differences in generic QoL between groups with and without SH (not further defined [30] or episodes requiring assistance from others [31]) in the past 6 [31] to 12 months [30].
Three studies explored the association between SH and diabetes-specific QoL using the DIS-ABKIDS DCGM-12 [13,28] or the DQOL-Y [33]. In two studies, SH (not defined [13] or episodes requiring assistance [28]) in the past 12 months was not associated with diabetes-specific QoL after adjustment for covariates (e.g., age, gender, HbA 1c ). However, one of these studies indicated that those who experienced SH (not further defined) in the past month reported significantly lower diabetes-specific QoL than those who had not [13]. The third study found that SH with coma since diagnosis was significantly associated with lower diabetes-specific QoL [33].
Non-severe hypoglycemia. One study found no significant differences diabetes-specific QoL scores on the DQOL-Y, between those who experienced NSH (glucose levels below 70 mg/dl) in the past three months and those who did not [32].
Two studies reported no significant association between frequency of NSH (glucose levels below 60 or 70 mg/dl) in the past month and generic QoL (PedsQL) after adjusting for covariates such as gender, hyperglycemia and age of onset of diabetes [16,29].
Broad and specific domains of QoL. Table 1 shows that studies examined three broad domains of QoL (psychological, physical, or social functioning), and/or the following specific domains of QoL: school, family, friends, self-esteem and sleep. Three studies used non-validated QoL questionnaires [13,14,34]. The participants' age range was 8-21 years, one study only included adolescents aged 12-18 [35].
Severe hypoglycemia. Four studies examined the relationship between SH (not further defined [13], or inability to self-treat due to neurological dysfunction [14,34] or requiring assistance from others [31]) and broad domains (physical, psychological, social), or specific domains (self-esteem, family, friends or school) of QoL.
Three studies found no significant relationship between SH in the past 1-12 month(s) and physical functioning, [13,14,34]. In contrast, one study indicated that those with two or more    episodes of SH in the past six months reported significantly lower physical functioning than those without SH, but not for those who only had one SH [31]. Three studies found no significant relationship between SH in the past six [31] to 12 months and psychological functioning [13,31,34]. However, one study examined various recall periods and found that SH in the past month was significantly associated with lower psychological functioning [13]. The fourth study reported that those with SH in the past three months reported significantly lower psychological functioning than those without [14]. Two studies examined the association between in the past 1-12 months and self-esteem and found no significant results [13,34].
One study reported no significant associations between SH in the past six months and social functioning [31]. Three studies reported no significant associations between SH in the past 1-12 months and relationships with family [13,14,34]. None of the studies reported a significant association between SH in the past 3-12 months and school [13,14,31,34], although one study indicated that SH in the past month was significantly associated with lower school functioning [13]. Three studies examined the association between SH and quality of friendship [13,14,34]. One found that SH in the past year was significantly associated with lower quality of friendship, but this was not observed for SH in the past month [13]. Two studies found no significant association between SH and quality of friendship in the past 3-12 months [14,34].
Non-severe hypoglycemia. Two studies examined associations of NSH and domains of QoL using the PedsQL [16] or the ASWS [35]. The first study found no significant associations between NSH (glucose levels below 70 mg/dl) in the past month and physical functioning or psychosocial functioning [16]. The second study found no association between nocturnal hypoglycemia (glucose levels below 70 mg/dl or symptomatic hypoglycemia) in the past month and adolescents' sleep quality [35].
Four studies examined relationships between SH and worries about hypoglycemia [36,38,41,42] Three of these studies reported significant, small-to-medium, positive correlations between frequency of SH (not further defined [41] or inability to self-treat due to mental disorientation or seizures [38,42]) in the past 12 months and greater worries about hypoglycemia. However, in one study, this only remained statistically significant for female adolescents after controlling for gender [38]. The fourth study found no significant difference in worries about hypoglycemia between adolescents who never lost consciousness and those who ever lost consciousness due to SH, after controlling for covariates such as age gender and other types of hypoglycemia [36].
Three of the studies also explored associations between SH and FoH related behaviors [36,38,41]. One study reported no significant association between frequencies of SH episodes (inability to self-treat due to mental disorientation or seizures) and FoH related behaviors [38]. The second study indicated that those who had passed out due to hypoglycemia significantly reported more hypoglycemia related avoidance behaviors compared to those who had never passed out [36]. In contrast, the third study reported that those who needed medical attention due to SH reported significantly less hypoglycemia related avoidance behaviors than those who did not [41].
Three of the eight studies examined associations between SH (inability to treat due to mental confusion or unconsciousness in the past three months [39] or SH resulting in seizures or coma [15,44]) and overall FoH, but found no significant associations [15,39,44].
An additional study found that frequency of SH (loss of consciousness or requirement of glucagon) in the past month was a significant predictor of self-reported post-traumatic stress (PTSD) assessed with the CPTS-RI after adjustment for age and family history of diabetes [47].
Non-severe hypoglycemia. Four studies explored the relationship between NSH and HFS-C subscale scores [36,38,41,42]. Three of these studies reported no significant associations between frequency of NSH (glucose values below 70 mg/dl [41] or interfering with ability to function [38,42]) and FoH. The fourth study found that frequency of NSH and 'hypoglycemia while at school', 'awake' or 'asleep' were significantly associated with at least one of the HFS-C scales, after adjustment for clinical factors and other types of hypoglycemia (e.g., passing out because of hypoglycemia). This was not observed for 'hypoglycemia in front of friends' [36].
Related outcomes-diabetes-specific. Table 1 shows that studies assessed the relationship between hypoglycemia worries attributed to diabetes, diabetes-related disordered eating and diabetes distress. One study used a non-validated questionnaire [34]. The participants' age range was 5-21 years. Two studies focused on adolescents aged 12-18 [30,42] and one study conducted analysis for children and adolescents separately [28].
Severe hypoglycemia. Eight studies investigated the association between SH and diabetes distress [13,15,28,30,34,37,42,45]. Four of these studies used the PedsQL DM [15,30,42,45]. One of these reported that SH (not further specified) in the past 12 months was significantly associated with higher diabetes distress, compared to those without SH, with a small effect size [30]. This was confirmed for those who experienced two or more SH episodes (requiring assistance from others) in the past six months, but not for only one SH [45]. In contrast the other two studies reported no significant association between SH (inability to selftreat due to mental confusion) in the past 12 months) [42] or SH (resulting in seizure or coma) since diagnosis [15] and diabetes distress.
Two of the studies using the DISABKIDS Diabetes Module found a significant association between SH (not further defined [13] or requiring assistance from others [28]) and greater diabetes distress if hypoglycemia was experienced in the past month [13] but not in the past year [13,28]. One study reported no significant association between SH (inability to self-treat due to neurological dysfunction) and illness-related distress using the KINDL-R [34].
An additional study (using the DQOL-Y) reported that those who had SH involving seizures or coma in the past three months reported significantly more worries about diabetes than those without [37].
In addition, in one study frequency of SH (requiring assistance from others) in the past 6 months did not significantly differ between those with and without disordered eating measured with the DEPS-R [50].
Non-severe hypoglycemia. Four studies examined the association between NSH (glucose concentrations below 60 or 70 mg/dl [29,32,46] or interfering with ability to function [42]) in the past 1-6 months and diabetes distress using the PedsQL DM [29,42,46] or the DQOL-Y [32]. None of these studies reported a significant association between NSH and diabetes distress.
Related outcomes-generic. Table 1 shows that generic outcomes including anxiety or depression symptoms, or diagnosis were examined in the studies. Some of the studies used measures that are not validated in young people with diabetes [36,38,43,44]. The participants' age range was 0-25 years.
Severe hypoglycemia. Three studies reported no significant association between SH and depressive symptoms; the first explored the association between SH in the past 6 months (not further defined) and CDI-S scores [48] and the other two explored the association between SH resulting in coma or seizure and CES-D scores [43,44]. In contrast, another study reported a significant positive relationship between SH (requiring assistance from others and unconsciousness or application of glucagon) in the past year and a DSM-IV depression diagnosis [49].
Two studies investigated the associations between SH and anxiety symptoms assessed by the SCARED [36] or an ICD-10 anxiety disorder diagnosis [40]. The first study reported that a history of passing out due to SH was significantly associated with greater symptoms of separation anxiety and school avoidance, but not with panic disorder, generalized anxiety or social anxiety [36]. The second study reported no significant associations between SH (loss of consciousness) and diagnosis of anxiety disorders [40].
Non-severe hypoglycemia. Two studies examining associations between hypoglycemia in different situations and various anxiety types (using the SCARED [36] or STAIC [38]) found that having 'hypoglycemia while at school', 'in front of strangers', 'while awake' or 'asleep' [36] was significantly associated with greater symptoms anxiety, for example social anxiety or separation anxiety [36] and that hypoglycemia in social situations was significantly associated with higher trait anxiety, with a moderate effect size [38].

Discussion
To our knowledge, this is the first systematic review that critically examines evidence on the relationship between hypoglycemia and QoL and related outcomes among children and adolescents with type 1 diabetes. Results of this review show that evidence regarding an association between SH and (domains of) generic QoL is inconclusive, while the evidence suggests no association between NSH and generic QoL. For diabetes-specific QoL, the evidence was too limited to draw conclusions. None of the studies used hypoglycemia-specific QoL measures to explore the association between hypoglycemia and QoL. In addition, there was some evidence suggesting an association between SH in the past 12 months and greater worries about hypoglycemia, and no association between NSH and diabetes distress. There was insufficient evidence to draw conclusions regarding the relationship between hypoglycemia and diabetes distress (for SH), FoH worries (for NSH), FoH-related behaviors and total FoH, anxiety, depression, disordered eating and PTSD.
A possible explanation for inconsistent findings is the heterogeneity in definitions of and recall periods for hypoglycemia and measures used to assess QoL across studies. This variation limits the ability to compare studies and draw conclusions. Several key limitations of the existing evidence base were identified, such as cross-sectional designs, low statistical power, lack of reporting of effect sizes (and thus limited information on the clinical value of the observed statistically significant differences), lack of information on the definition or frequency of hypoglycemia, and the self-report of hypoglycemia over several months or even back to diagnosis, which might have led to recall bias. The key recommendation for future studies is to use a definition of hypoglycemia as recommended by current guidelines. Future studies should also use longitudinal /prospective study designs and modern methods, such as continuous glucose monitoring, for a more objective assessment of hypoglycemia that does not rely on recall of episodes, to determine the direct, day-to-day impact of hypoglycemia on various domains of QoL in children and adolescents with type 1 diabetes.
Although the current evidence suggests no clear association between hypoglycemia and some outcomes, it is important to note that studies were more likely to show statistically significant associations between hypoglycemia and outcomes when SH was experienced recently (in the past 1-3 months), more frequently, or when it involved convulsions, unconsciousness, or coma. In addition, some studies suggested that the context in which hypoglycemia takes places (e.g., in social situations) might have implications for its impact. However, this was only based on a few studies, some of which have methodological limitations. Thus, more evidence is needed to confirm these associations.
Although emerging evidence shows the importance of self-treated hypoglycemia in relation to QoL and related outcomes in adults with diabetes (50,51), current evidence on this relationship in youth with type 1 diabetes suggested no association between NSH and QoL. However, this should be interpreted with caution, as these studies are limited by the use of generic and diabetes-specific QoL questionnaires, while hypoglycemia-specific QoL measures may be more sensitive to the impact of NSH. Different research designs, that minimize recall bias and assess the impact closer to the occurrence NSH are needed to understand the association between NSH and QoL. In addition, this review identified only one study that explored the association between hypoglycemia while asleep and sleep quality (28). This highlights the need for more studies that investigate such relationships.
Although QoL has been considered as a key outcome in pediatric diabetes care [12], only two studies had a primary aim to examine the impact of hypoglycemia on QoL [15,42]. Fifteen of the 27 studies aimed to explore QoL, however, only seven of these studies included measures that actually assess QoL, whereas the others focused on particular domains of QoL or measured related outcomes such as diabetes distress [13,15,28,30,34,37,42,45] or health status [14], rather than QoL. Even though other studies included in this review have focused on identifying sociodemographic and clinical factors that are associated with QoL, it is difficult to identify the impact of hypoglycemia specifically in these studies [13, 14, 16, 28, 30-34, 37, 45, 46]. Importantly, some of the studies that explored the impact of hypoglycemia on QoL as a secondary aim had very low rates of hypoglycemia in their samples. To truly understand the impact of hypoglycemia on QoL, a questionnaire that assesses how hypoglycemia affects domains of life that are important to the individual should be used [18]. Given that there are currently no hypoglycemia-specific QoL measures that are designed to assess the impact of hypoglycemia on QoL in children and adolescents with type 1 diabetes, these need to be developed and would need to be age appropriate and to incorporate specific domains that are important to young people with diabetes. There might be other domains of importance to young people's QoL, such as leisure activities, that were not included in questionnaires used in current studies.
Finally, the majority of studies included in this review pooled children and adolescents together when examining the link between hypoglycemia and QoL or related outcomes. Although these studies usually included age-appropriate assessments of outcomes, the impact of hypoglycemia on these outcomes might be different for children and adolescents. While younger children often rely on their parents for decisions about diabetes management, these responsibilities are usually transferred to the child during adolescence [53,54]. During this challenging process of transferring responsibilities, the burden of self-management for the adolescents increases and can lead to increased hypoglycemia, which can interfere with other demands and lead to family conflicts, reduced self-efficacy and increased FoH, all aspects that can compromise QoL in adolescents with diabetes [42,55]. Future studies should thus explore if the relationship between hypoglycemia and QoL is different in different age groups. Additionally, adolescence is characterized by a strong desire to be accepted by peers [54]. Episodes of hypoglycemia in this age group could be experienced as embarrassing and mark out adolescents with diabetes as different. Hypoglycemia has indeed previously been associated with higher stigma in young people with diabetes [56], which can lead to poorer psychosocial and medical outcomes [57]. Future studies need to explore the role of stigma as a possible mechanism by which hypoglycemia impacts on QoL.

Strengths and limitations
Strengths of this review include the systematic and comprehensive search of multiple databases, and the application of a conceptual framework of QoL to categorize outcome measures in order to critically appraise the evidence. This allows for a more detailed understanding of the various ways in which hypoglycemia can impact on QoL and related outcomes and highlights the gaps in the evidence base. This review also has some limitations. Although the inclusion of a wide range of outcomes provided an overview of all the available evidence related to the impact of hypoglycemia on QoL, it also made it difficult to compare studies directly. Further, the heterogeneity across studies and the lack of effect sizes reported in studies, precluded the possibility of meta-analysis. The inclusion of only quantitative studies that were published in English may have introduced some bias, although only six studies were excluded for this reason.

Implications for clinical practice
The implications for clinical practice that can be drawn from this review are limited due to the inconclusive and relatively small evidence-base. However, some evidence suggests that more recent episodes of hypoglycemia might have an impact on various outcomes. This may be useful for clinicians, as they could ask specifically about hypoglycemia and its impact in the weeks/months following episodes of SH.

Conclusion
This systematic review shows that there is insufficient evidence on the relationship between hypoglycemia and (domains of) generic and diabetes-specific QoL in children and adolescents with type 1 diabetes. This is largely because heterogeneity and methodological limitations across studies hamper the ability to draw strong conclusions. Importantly, none of the studies used a measure designed specifically to assess the impact of hypoglycemia on QoL. Additionally, there seems to be an association between SH and greater worry about hypoglycemia, while the evidence is too limited for other related outcomes. Although limited, some evidence suggests that issues such as timing and context of hypoglycemia might influence its impact. Future research should focus on the development of measures that can assess the impact of hypoglycemia in children and adolescents with type 1 diabetes and use agreed definitions of hypoglycemia that increase comparability between studies.