A systematic review of grandparents’ influence on grandchildren’s cancer risk factors

Many lifestyle patterns are established when children are young. Research has focused on the potential role of parents as a risk factor for non communicable disease in children, but there is limited investigation of the role of other caregivers, such as grandparents. The aim of this review was to identify and synthesise evidence for any influence grandparents’ care practices may have on their grandchildren’s long term cancer risk factors. A systematic review was carried out with searches across four databases (MEDLINE, Embase, Web of Science, PsycINFO) as well as searches of reference lists and citing articles, and Google Scholar. Search terms were based on six areas of risk that family care could potentially influence–weight, diet, physical activity, tobacco, alcohol and sun exposure. All study designs were included, as were studies that provided an indication of the interaction of grandparents with their grandchildren. Studies were excluded if grandparents were primary caregivers and if children had serious health conditions. Study quality was assessed using National Institute for Health and Care Excellence checklists. Grandparent impact was categorised as beneficial, adverse, mixed or as having no impact. Due to study heterogeneity a meta-analysis was not possible. Qualitative studies underwent a thematic synthesis of their results. Results from all included studies indicated that there was a sufficient evidence base for weight, diet, physical activity and tobacco studies to draw conclusions about grandparents’ influence. One study examined alcohol and no studies examined sun exposure. Evidence indicated that, overall, grandparents had an adverse impact on their grandchildren’s cancer risk factors. The theoretical work in the included studies was limited. Theoretically underpinned interventions designed to reduce these risk factors must consider grandparents’ role, as well as parents’, and be evaluated robustly to inform the evidence base further.


Introduction
Many lifestyle patterns are established when children are young. This is especially true for diet and physical activity patterns [1,2], two lifestyle areas where there is substantial evidence for the important influence of parental and other role models [3][4][5][6][7]. For a practice such as smoking, it is during the teenage years that this habit is generally established [8]. Smoking, diet and physical activity, along with excess weight, have been identified as risk factors for non communicable disease, particularly cancer [9]. Cancer is the leading cause of death in many countries in Western Europe [10], however, up to 40% of cancer cases could potentially be prevented through healthier lifestyles [9]. Research evidence on cancer prevention is limited compared with that of cancer treatment, however, there is evidence to suggest that exposure to risk factors in childhood increases an individual's likelihood of cancer morbidity or mortality in adulthood [11][12][13][14][15][16].
Factors associated with children's long term cancer risk are first experienced within the family setting. The nuclear family of a father, mother and their children, has been the dominant family model in Western societies in more recent generations [17]. Nevertheless, changes in social conditions, such as an increase in lone parenting, more women in the workforce and prohibitive childcare costs, have led to an increased focus on the role of grandparents' in their grandchildren's lives. Health improvements have resulted in greater life expectancy enabling grandparents to support their families by providing childcare, or spending more time with their grandchildren as they are growing up.
Around one fifth of 0-12 year olds in Australia [18], and a quarter of pre-school children in the US [19], are regularly cared for by grandparents. In the UK, it is estimated that grandparent care saves parents around £1700bn per year in childcare costs [20]. The extent of grandparent involvement can vary based on cultural and societal differences. For example, when comparing childcare practices across Europe, Hank and Buber [21] found that grandparents in Greece, Italy and Spain were more likely to provide regular childcare to their grandchildren, and that grandparents in the Netherlands, France and Nordic countries were least likely to. These patterns reflect the differing social and cultural contexts in these countries, such as labour market participation by older women and state provision of formal childcare [22]. Within the UK, the important childcare role that grandparents provide has been recognised at government level with grandparents caring for grandchildren entitled to receive National Insurance Credits towards their state-provided pension [23]. Forthcoming legislation will also allow grandparents to share parental leave with parents in a child's first year of life, and for employed grandparents to have the right to work flexibly to allow them to care for their grandchildren [24].
There is a significant literature around the impact of caring for grandchildren on grandparents' health, particularly when grandparents are called upon to become primary carers to grandchildren. While some studies indicate that caring for grandchildren can have an adverse impact on grandparents' health [25][26][27], there is evidence that after controlling for sociodemographic factors, this caring role can have a beneficial impact on physical health [28]. It is less clear how grandparents' care influence their grandchildren's health. While there is some evidence that grandparents can play a significant role in supporting their grandchildren's social and emotional wellbeing [29,30], there is also evidence that the multifaceted nature of these relationships can have both beneficial and adverse impacts [31,32].
The influence of grandparents' care practices on grandchildren's physical health is also unclear, particularly for non-communicable diseases like cancer which are more likely to be experienced later in their grandchildren's lives. With greater recognition of the key role of grandparents in grandchildren's lives, there have been calls for parenting advice to be broadened to encompass grandparents. For example, in Scotland, this has been proposed as a potential strategy to help tackle health and educational inequalities in communities with high levels of disadvantage [33]. There is therefore a need to identify, review and synthesise the literature on grandparents' influence on their grandchildren to inform practitioners, policy makers and academics further about family dynamics that impact on health outcomes.
The aim of this systematic review was to identify and synthesise evidence for the influence of grandparents on their grandchildren's long term cancer risk factors. Key objectives were: 1. To examine the availability of evidence for grandparents' influence on a range of grandchildren's cancer risk factors; 2. To identify whether this evidence indicates that grandparents have a beneficial or adverse impact on grandchildren's cancer risk factors.
3. To identify theoretical frameworks used to inform research in this area.

Methods
The breadth of factors considered in the review was determined via Cancer Research UK's research on key preventable risk factors for cancer [9]. Cancer risk factors applicable to children that could potentially be influenced by grandparents were selected. These included tobacco (smoking or exposure), obesity, diet, alcohol consumption, sun exposure and inactivity. PRISMA guidelines were followed in reporting this review [34].

Search strategy
Searches were carried out using defined terms in Web of Science, Medline, Embase and Psy-cInfo from database start dates to May 2017. The research team hand searched the reference lists of included papers, and citing papers, and carried out a search of Google Scholar using variations of the search terms in S1 Table. Search terms Synonyms were identified around the Population (grandparents and grandchildren) and Outcome (cancer risk factors) components of the PICOS framework (see S1 Table). MeSH terms and subject headings were used where appropriate. The Boolean operator OR was used to combine within Population and Outcome search results, with AND used to combine these two blocks, to search titles and abstracts.

Selection of articles
All study years and designs were included if the relevant risk factors were examined. A further inclusion criterion was that publications must provide an indication that grandparents spent time with their grandchildren, either by providing childcare, living with children or during visits. Exclusion criteria included grandparents who acted as primary caregivers, grandchildren with serious medical conditions, and studies examining grandparents' health outcomes only. Abstracts, newspaper reports and non-English language publications were also excluded.

Quality assessment
Study quality was assessed using National Institute for Health and Care Excellence checklists relevant to the particular study design [35]. For quantitative studies (including observational, cohort and intervention studies), quality assessment focused around the representativeness of the study population, the method of selection, reliability of outcomes, and appropriate analyses. For qualitative studies, assessment was made based on aim, design, data collection, researcher role, description of context, appropriate methods and analysis, richness of findings and conclusions. Assessment of review articles was based on a focused question, relevance of included studies, rigour of search, study assessment and appropriately described methods. An overall study quality measure was also provided (high, medium or low) based on scores for the individual components assessed in each of the studies.

Data extraction
Data was extracted using a predefined form adapted from the Cochrane Collaboration [36]. Extracted data included study geography, participant characteristics, sample size, study aim, theoretical framework, outcome measures, main findings, and for qualitative data, participant quotations and author syntheses that discussed grandparent impact on grandchildren for the relevant risk factors.

Data synthesis
Given the range of potential outcome measures in the included quantitative studies, metaanalysis of data was not carried out. Instead, grandparent impact was defined as beneficial, adverse, mixed (some beneficial and some adverse outcomes) or as having no impact for all study types. Qualitative data then underwent further synthesis through a thematic analysis of both participants' quotations and article syntheses. Extracted text was read and reread by two reviewers. They each identified key themes and shared these with all study authors. Final themes were then agreed upon, and text coded under each theme. Themes were reorganised and structured hierarchically where possible.

Results
The searches yielded 5745 publications after removal of duplicates (Fig 1). All titles and abstracts were screened by two reviewers, with a third reviewer providing advice when disagreements on inclusion arose. This resulted in 134 publications retrieved for full text inspection, and 44 included in the analysis. A further 12 were retrieved from reference list and Google Scholar searches. A total of 56 publications were included. Explanations for exclusion of studies at full text stage were no indication of grandparents spending time with their grandchildren, or grandparents being primary caregivers, and no focus on children's cancer risk factors.
Study aims. Nineteen studies had a specific aim of investigating the influence of grandparents on their grandchildren's diet, physical activity or weight outcomes [32,52,58,61,63,65,66,68,71,74,75,77,81,83,[85][86][87][88]90]. These studies included a mix of qualitative and quantitative approaches. Three studies were literature reviews. One aimed to provide an overview of the evidence on interventions to reduce overweight and obesity in children within China [53]. The second aimed to synthesise qualitative literature on parental perceptions around weight-related behaviours [60]. The third aimed to gather and synthesise research findings on the effects of grandparent involvement on children's physical health outcomes [32]. Other studies' aims included investigating a range of factors (with grandparents included as one of a number) that might impact on children's weight or weight-related behaviours, and were all quantitative [55,57,59,62,67,76]. The remaining studies were qualitative and sought to gain a general understanding or additional insight into general weight, diet and food related issues in children [54, 56, 64, 69, 70, 72-74, 78-80, 82, 84, 89].
Theoretical framework. Five studies outlined or used a theoretical framework to guide and/or analyse their investigation. Goh et al. [88] used the Intergenerational Parenting Coalition approach to guide the study, that is the recognition that three-generational living forms part of China's embedded cultural context, and that the interactions between generations are dialectical in influence. Sowan and Stember [62] drew on the Web of Causation Model [91] to examine how parental characteristics can impact children's risk of obesity. This model focuses on risk and the interrelationships between risk factors. Styles et al. [82] used the socio-ecological approach [92] (the different levels that impact on an individuals' health practices) to examine parents' and grandparents' concerns around obesity prevention, classifying them around intrapersonal, interpersonal, organizational and environmental levels. Boni [84] drew on Practice Theory in an ethnographic study that examined children's food cultures in post-Soviet Poland, whilst Eli et al. [71] used a conceptual framework of familial homeostasis to discuss intergenerational feeding of children.
Study results. Weight-The evidence was strongest for grandparents having an adverse impact on children's weight outcomes. No studies found a solely beneficial impact. Eight  To explore the personal relationship between a mother and her mother, grandmother or other female relation and its influence on the present family's food choices.  studies of the 17 studies found grandparents to have an adverse impact. Three studies found a mixed impact [32,61,64], and four found grandparents to have both adverse or no impacts for various relevant outcomes [58,[66][67][68]. Two studies found no impact [55,62]. Diet-Similar to weight, grandparents overall appeared to have an adverse impact on their grandchildren's diets [69, 70, 72, 73, 76, 78-80, 85, 88, 89], with an additional four studies reporting both adverse/no impacts [58,65,66,83]. Nine studies reported mixed impacts [32,71,74,75,81,82,84,86,87]. One study found mixed/ no impact of grandparents [61]. Kaplan et al. [77] was the only study to find a solely beneficial impact of grandparents in managing and encouraging healthy eating in their grandchildren. Physical activity-The evidence relating to children's activity was less conclusive than for diet and weight, however, there was still greater evidence that grandparents' actions had an adverse impact on grandchildren's outcomes. Four studies identified adverse outcomes [52,58,76,79], one study found mixed outcomes [71], one adverse/no impact [58], and a single study found no impact [65]. Lako [90] was the only study to find beneficial outcomes. They described both grandparents' practical and support roles in transporting grandchildren to sporting activities, and actively cheering them on.
Thematic synthesis. Thematic analysis of extracted weight related qualitative data identified two broad themes describing grandparent impact: 1) Influence on family relationships and 2) Grandchildren's diet. A more specific theme on physical activity was also identified. All studies reported some adverse impacts, with parents and grandparents reporting divergent views on appropriate eating behaviour [54, 56, 58, 64, 69, 70, 73-75, 78-80, 82, 84, 85, 88, 89]. This included the type of food provided, for example, high sugar or fat foods, or providing too much food. Parents reported feeling frustrated and undermined, and described these practices  as 'spoiling' grandchildren. The need to rely on grandparents for childcare often resulted in grandparents' practices prevailing [73,75], however in other instances, grandparents reported that they followed parents' rules. Johnson et al. [75] and Eli et al. [71] reported both adverse (eg, disconnected, ambivalent) and beneficial influences on family relationships. Where relationships were disconnected or ambivalent, mothers described rejecting some or all of the food practices that were modelled by grandmothers. These parents wished to provide their children with healthier foods, or to change mealtime practices, such as the rule that children must clear their plate. Conversely grandparents could also reject parents' healthier food practices. Beneficial relationships were described through the intergenerational transmission of cooking skills and practices, but also through grandmothers being mindful of parents' wish for children to eat healthily.
Four different roles around grandchildren's diets were identified for grandparents', (1) Buying and preparing food 2) Excessive and non-recommended feeding 3) Food as control/love 4) Promoting healthy food choices. 1) Buying and preparing food-A number of studies described grandparents as a source of support for parents either in buying, preparing and cooking food [71,72,74,86,88]. While the preparation of meals from scratch with fresh ingredients could be seen as a positive, this was undermined by the role grandparents could play in overfeeding children or feeding less healthy foods [71,74,79,82,85]. 2) Excessive and nonrecommended feeding-Jiang et al. [74] and Li et al. [58] described grandparents' excessively feeding children as a form of nurturing, where grandparents believed that overfeeding and excess weight were signs of health, and that this was a response to poverty and hunger experienced by grandparents in their youth. The feeding of non-recommended foods was demonstrated by grandmothers in Egypt, where they believed that children needed to eat 'light', sugary foods to thrive [78]. 3) Food as control/love-Related to overfeeding was grandparents using food to demonstrate their love for their grandchild [71,74,84]. This included practices such as physically feeding children who were capable of carrying this out for themselves, or providing foods prohibited by parents. Grandparents also said they used food a means through which to control grandchildren's behaviour and to reward them for achievements. Roberts and Pettigrew [80] found that Australian parents reported that grandparents provided unhealthy food as a strategy to create a stronger bond between them and their grandchildren. Strategies to reduce tensions between parents and grandparents were not discussed in any studies. 4) Promoting healthy food choices-This theme was described in two studies [75,77], and involved grandparents engaging with children in interactive ways to promote healthy eating, such as through humour, or by involving them in meal planning and preparation.
A final theme was grandparents' impact on children's activity-related practices. These included, physical activity, sedentary behaviour and screen time, and were discussed in three studies [58,71,90]. There was no overall direction in which grandparents appeared to impact these practices. Some grandparents put limits on children's screen time, whist others allowed the same access as that provided at home. Physical activity levels appeared to be related to whether grandparents were active themselves, or whether there was appropriate space where children could be active. Grandparents were supportive of children's participation in physical activity, and often enabled it through facilitating children's access to spaces in which they could carry this out. In contrast to food, there appeared to be less tension in relation to parenting practices around activity.
Study aims. Four studies examined the impact of grandparents on children's smoking, or their exposure to secondhand smoke [37,[49][50][51]. Other studies aimed to examine the relationship between children's levels of SHS exposure [38-41, 44, 47, 52] or respiratory ill health [39,42] and a range of possible environmental exposures (including grandparents), or potential influences on children's smoking behaviour [43,46,48]. The intervention study evaluated new methods for nurses to support parents in protecting their children from SHS [45].
Theoretical framework. Four studies made reference to theory. Robinson et al. [47] analysed in-depth interview data drawing from aspects of social theory, though this was not specified further. Escario and Wilkinson and Duarte et al. (using data from the same survey) used social learning theory to frame their cross-sectional analyses of in-home influences on children's smoking practices. Mao [50] used theories of gender inequality to investigate the role of mothers' and grandmothers' in regulating grandfathers' smoking around young children.
Study results. Nine studies out of 16 found that there was an adverse impact of grandparents on children's smoking or exposure to SHS [38, 39, 41, 44-46, 48, 51, 52]. Three additional studies found a mix of adverse impacts for some outcomes and no impact for others [40,42,43]. Living with a grandparent who smoked was problematic [37,40,42,[49][50][51], as was visiting grandparents who smoked [38,39,41,44,45]. Mao [50] found that living with grandfathers who smoked increased children's secondhand smoke exposure, but that living with grandmothers was protective. Profe and Wild [37] found no impact of grandparent involvement on smoking. Robinson et al. [47] identified a beneficial impact of grandparents, with participants reporting that becoming a grandparent had prompted grandparent smokers to reassess their habits, and to no longer smoke indoors when their grandchildren were present. Escario and Wilkinson [49] found that living with a grandparent reduced the likelihood that a child would smoke, but had no impact on the consumption levels of children who did smoke.
Study quality (see S5-S7 Tables). For quantitative tobacco studies, three were rated as high quality [40,42,46], four were rated as medium quality [44,49,51,52] and four as low [37,38,41,43]. Low quality studies had low sample sizes or unrepresentative samples and/or did not control adequately for confounding. Two qualitative studies were rated as high quality [39,47], and two medium quality [48,50]. The single intervention study (a before/after design) was rated as medium quality [45], with weaknesses explicitly around analysis.
Thematic synthesis. The four qualitative studies provided only limited data for the thematic synthesis [39,47,48,50]. The three themes were, 1) parents limiting grandparent access to grandchildren who smoke, 2) grandparents protecting children from SHS, and 3) grandparents acting as negative role models. For the first theme, parents described their inability to enforce rules around grandparents smoking in the home, and that this resulted in parents limiting grandparent interaction with grandchildren [39,50]. In contrast Robinson et al. [47], found that the birth of grandchildren was a catalyst that encouraged grandparents to stop smoking completely, or to stop smoking indoors when their grandchildren were present. This was also true of some grandparents in Mao's study of Chinese grandparents [50]. The third theme of grandparents acting as negative role models was discussed by Thai women [48] who described growing up around grandparents who smoked. As well as seeing smoking practices frequently, they became more involved through buying or rolling tobacco for grandparents. Women said they believed this early exposure and involvement in smoking practices partly explained their own smoking as an adult.

Alcohol
A single study examined the impact of grandparents on children's alcohol consumption (Table 4). Prof and Wilde [37] used cross-sectional data gathered from adolescents in South Africa to investigate whether grandparent involvement predicted use of alcohol. The results found no significant impact, and the study was rated as low quality.

Discussion
This review has been the first to identify and synthesise evidence for the influence of grandparents on their grandchildren's long term cancer risk factors. Results indicated that there was a sufficient evidence base for weight, diet, physical activity and tobacco studies to draw conclusions about grandparents' impact. There was minimal evidence for alcohol and no evidence for sun exposure.
Studies that showed a beneficial impact highlighted that grandparents did not always undermine parents, and could play a role in promoting healthy eating practices [32,57,61,64,65,71,75,77,81,82,84,86,87,90]. Robinson et al. [47] provided a high quality in-depth study of smokers, ex-smokers and their families, identifying grandparenthood as a pivotal point for behaviour change, either by deciding to stop smoking completely or stop smoking in the home. It was not possible to identify under what circumstances these beneficial impacts took place due to the heterogeneity of the included studies.
The results indicate a lack of theoretical rigour in most of the studies in this area. Only nine studies used or made reference to an explicit theoretical framework. These included the intergenerational parenting coalition [88], web of causation [62], socio-ecological models [82], social learning theory [49,51], practice theory [84], familial homeostasis [71] and theories of gender inequality [50]. In identifying grandparents as impacting adversely on their grandchildren's cancer risk factors, studies failed to take into account the wider context in which the results are to be understood. Grandparents are likely to be one of many influences on health outcomes, and are located at the interpersonal level of the socio-ecological impacts on health [92]. Indeed grandparenting exists within a complex social system in which it interacts with influences at a variety of levels in children's lives. Few of the studies above discuss these in depth, but they suggest some potentially important influences, such as parents' working patterns, societal norms and lack of other childcare options.
The studies included in the review do not take into account the more general beneficial role grandparents may play in their grandchildren's lives. Grandparents may be better able to spend time with their grandchildren in ways that parents are unable to. This can help facilitate good social and emotional wellbeing in grandchildren, and therefore, any recommendation to limit grandparent interaction with their grandchildren would be misplaced. Instead, as suggested by results from the Growing Up in Scotland cohort studies [33], parenting advice and support needs to be broadened to encompass grandparents as well as parents. Grandparents' roles must be recognised and practical steps put in place to facilitate optimal intergenerational parenting. In some studies, there were hints that tensions could arise between parents and grandparents, with little suggestion of how communication between generations could be enhanced to ensure that shared understandings around parenting could be realised. In addition, there were no interventions identified that sought to encompass grandparents as a potential mechanism through which to improve grandchildren's diets. With the caring role of grandparents now being recognised within the UK legislation and benefits system, and the expectation that grandparents' involvement in their grandchildren's lives will only increase, there is a need for theoretically grounded interventions to be designed that include significant communication-based components.

Strengths and limitations
This study has integrated the evidence-base on the impact of grandparents on grandchildren's cancer risk factors. Cancer research has focused more on treatment of disease rather than the full range of factors that might play a role in cancer prevention over the life course. The review therefore took a broad approach to the types of evidence considered for syntheses, with qualitative literature synthesised through a thematic analysis of participant quotations and author analyses. A thorough quality appraisal also took place using appropriate tools for each of the study types included. A larger proportion of qualitative studies were rated as lower quality compared with quantitative studies.
While the review used a range of key databases to identify relevant articles, it did not ask authors or experts to identify additional studies, and did not include findings from the grey literature. In addition, non-English language studies were not included, which limits the applicability of the review findings across cultures. An additional limitation was that many studies contained only a limited description of grandparents' impact, and/or provided little indication of the extent to which the amount of time grandchildren spent with grandparents was associated with more adverse outcomes or behaviours.

Conclusion
The weight of the evidence within this review found that grandparents had an adverse impact on their grandchildren's cancer risk factors. Future work should focus on realising the potential for grandparents to be a positive influence on their grandchildren's health through the design of realistic, theoretically underpinned interventions. Interventions should ideally include components that aid facilitating family communication around areas of tension. The formative stages of this work should include the perspectives of both grandparents and grandchildren to enhance the likelihood of success.
Supporting information S1  Table. Observational and cohort study quality-weight, diet and physical activity studies. ++ Indicates that for that particular aspect of study design, the study has been designed or conducted in such a way as to minimise the risk of bias. + Indicates that either the answer to the checklist question is not clear from the way the study is reported, or that the study may not have addressed all potential sources of bias for that particular aspect of study design.-Should be reserved for those aspects of the study design in which significant sources of bias may persist. NR-Not reported-Should be reserved for those aspects in which the study under review fails to report how they have (or might have) been considered. NA-Not applicable-Should be reserved for those study design aspects that are not applicable given the study design under review. (DOCX) S3  Table. Observational and cohort study quality-tobacco studies. ++ Indicates that for that particular aspect of study design, the study has been designed or conducted in such a way as to minimise the risk of bias. + Indicates that either the answer to the checklist question is not clear from the way the study is reported, or that the study may not have addressed all potential sources of bias for that particular aspect of study design.-Should be reserved for those aspects of the study design in which significant sources of bias may persist. NR-Not reported-Should be reserved for those aspects in which the study under review fails to report how they have (or might have) been considered. NA-Not applicable-Should be reserved for those study design aspects that are not applicable given the study design under review. (DOCX) S6