Factors that influence the health of older widows and widowers—A systematic review of quantitative research

Abstract Aim To examine factors that influence the health of older widows and widowers. The review question was: What is the evidence of the relationship between widowhood and health in older adults? Design Systematic review. Data sources Academic Search Elite, CINAHL, Medline (Ovid) and PubMed were searched for articles published between January 2013–December 2017. Review methods A systematic review of quantitative research with a qualitative thematic analysis. Results The selection process resulted in 12 studies. One of the themes that emerged was: emotional challenges related to experiences of bereavement, depression and anxiety, which was based on the sub‐theme social support as the main strategy for coping with emotional pain and suffering. The second theme was: struggling with poor physical health. The findings indicate that healthcare professionals need knowledge and skills to deal with the health consequences of widowhood in old age. Building community teams can prevent emotional and physical health problems, as well as reduce mortality.

that having no children is significantly associated with depression among older widowed adults. Depression has been related to expectations and burden of support as well as being a care recipient (Tiedt, 2010(Tiedt, , 2013Tiedt, Saito, & Crimmins, 2016). Zhang and Li (2011) revealed the effect of marital status on depressive symptoms that was mediated by family support and moderated by the support of friends. Research from Mexico found that social integration can both mitigate and exacerbate depression among older widowed adults (Monserud & Wong, 2015). Thus, as mentioned by Aniruddha (2013), widowhood seems to have mental, social, behavioural and biological consequences, consistent with a stress-inducing process.
The meaning of bereavement, grief and mourning seems to differ across cultures, where most societies outline appropriate behaviours for those who are widowed based on socially constructed sets of norms (Robben, 2018). Cultural factors seem to influence health, and Lloyd-Sherlock, Corso, and Minicuci (2015) found variations in the prevalence and timing of widowhood across countries such as China, Ghana, India, Russia and South Africa, in addition to associations between widowhood and being in the poorest wealth quintile of these countries. However, the evidence of the difference in impact across regions related to the cultural implications of widowhood on both individual and societal level is unclear. Widowhood is described as a cultural and gendered experience because the salience of different mechanisms linking widowhood to health may depend on local norms (Uhlenberg, 2009). Widowhood seems to trigger various health problems (Shear et al., 2011). The experience of losing a spouse appears to change over time (Williams et al., 2012).
Three systematic reviews (Holm & Severinsson, 2012;Lobb et al., 2010;Stahl & Schulz, 2014), one integrative review (Naef, Ward, Mahrer-Imhof, & Grande, 2013) and two reviews (Merz & De Jong Gierveld, 2016;Nseir & Larkey, 2013), focused on different physical and/or mental health factors associated with bereavement in older widowed adults. Holm and Severinsson (2012) reviewed evidence about the emotional state of older widows. Lobb et al. (2010) aimed to clarify current knowledge to inform future planning and work in the area of complicated grief following bereavement. Stahl and Schulz (2014) examined the relationship between late-life spousal bereavement and changes in routine health behaviours. Naef et al. (2013) determined key characteristics of the bereavement experience of older widowed persons. Nseir and Larkey (2013) examined the effect of interventions on the grieving process of older bereaved spouses. Merz and De Jong Gierveld (2016) investigated the role of family relationships through the lifespan in reducing loneliness among ever-widowed older adults (i.e., persons who experienced the death of a spouse at some time during their life). These different reviews identified important factors that influence aspects of physical health (Stahl & Schulz, 2014); identity, mental health (Holm & Severinsson, 2012;Lobb et al., 2010;Naef et al., 2013;Nseir & Larkey, 2013); and family relationships (Merz & De Jong Gierveld, 2016). The reviews provided comprehensive evidence of factors that influence the health of older widows and widowers around the world. Such factors seem to have undergone a transformational change in recent years, especially in relation to the understanding of the human experience of loss as pointed out by Hall (2014). Older adults mostly describe the loss of a spouse as causing grief that decreases over time (Bonanno, Moskowitz, Papa, & Folkman, 2005).
Despite our best efforts, we could find no review study examining factors that influence the health of older widows and widowers.

| AIMS
The aim of this systematic review was to examine factors that influence the health of older widows and widowers. The review question was: What is the evidence of the relationship between widowhood and health in older adults?

| Design
A systematic review method was used to explore the available evidence (Moher, Liberati, Tetzlaff, Altman, & The PRISMA Group, 2009). Systematic reviews have a set of objectives that can be used to reproduce methodology including a systematic search to identify studies that meet the inclusion criteria. In addition, it is necessary to assess the validity of the findings and present a synthesis of the characteristics of the included studies (Moher et al., 2009).

| Search strategy
Electronic searches were performed in Academic Search Premier (1), CINAHL (10), PubMed (191) and ProQuest (417)  The abstracts included studies that did not fulfil the inclusion criteria. The time period for this systematic review was limited to 5 years. The retrieval and selection process resulted in a total of 12 quantitative studies, presented in Figure 1.

| Inclusion and exclusion criteria
Inclusion criteria were as follows: studies published between 2013-2017, older adults aged 60 years and over, and published in the English language in peer-reviewed journals. Articles were selected based on the titles and keywords including concepts such as health, well-being, mental health, depression and bereavement. As most relevant studies were quantitative, the authors decided to include only quantitative studies. Exclusion criteria were as follows: review studies, theoretical studies, younger adults and published before 2013.

| Thematic analysis and synthesis
According to Holopainen, Hakulinen-Viitanen, and Tossavainen (2008), the analysis of a systematic review can be either statistical or qualitative, depending on the material and purpose of the study. A thematic analysis was used to synthesize evidence related to the experience of bereavement and consequences for the health of widows and widowers (Bornbaum, Koras, Peirson, & Rosella, 2015;Guest, MacQueen, & Namey, 2012). Data were extracted and analysed by the first author (ALH), who read the extracts several times to identify relevant descriptions of the experience of bereavement and its consequences for the health of older widows and widowers. Themes were identified by inductive coding. The extracted data were synthesized by ALH. The draft synthesis was reviewed by the second (AKB) and third author (ES) and refined until agreement on the themes was achieved. Data that covered the same issue were coded and grouped together until convincing descriptions of the themes emerged. The process of determining suitable formulations was time-consuming because the data were similar (Sandelowski, 2010).

| Quality appraisal and methodological characteristics
Five studies were described as cross-sectional and seven as cohort studies (Table 1). The studies were evaluated in accordance with the Critical Appraisal Skills Programme (CASP, 2014) (  (Panagiotopoulos, Walker, & Luszcz, 2013). Old baseline data can be a limitation and lead to a risk of bias, thus threatening validity, reliability (Polit & Beck, 2012) and distorting the results. Such data could reflect general and not specific individual differences (Spahni, Morselli, Perrig-Chiello, & Bennett, 2015). This could be related to the history threat described by Polit and Beck (2012) as a criterion of internal validity associated with the time factor, which means that the data may have F I G U R E 1 Flow chart of search outcomes and selection process of articles on the health status of widows and widowers (Moher et al., 2009)

1.
Was the aim/problem in the study clearly defined? 2.
Was the cross-sectional design a suitable method to answer the aim and research question(s)?
3. Was the population from which the sample was drawn clearly defined?

4.
Was the sampling method adequate?

5.
Was it explained whether (and how) the participants who agreed to participate are different from those who refused?

6.
Was the response rate adequate?

7.
Were the measurements shown to be valid and reliable?

8.
Were the procedures for data collection standardized?
The conclusion of the studies (not included in this table) 11. Can the results be transferred to practice?

12.
Do the results from this study support previous studies?

1.
Did the study address a clearly focused issue?

2.
Was the cohort recruited in an acceptable way?

3.
Was the exposure accurately measured to minimize bias?

4.
Was the outcome accurately measured to minimize bias?

5.
Have the authors identified all important confounding factors in the design and/or analysis? changed since collection. The cross-sectional studies used self-report questionnaires (Table 1). However, these studies contain no discussion about how self-report questionnaires might have increased response bias (Polit & Beck, 2012). Three studies provided no information about the response rate (Table 3), which can also increase the risk of bias. Two of the cross-sectional studies have methodological recommendations about future research (Table 1). In two of the studies, the use of a longitudinal design in future research is recommended (Perkins et al., 2016;Spahni et al., 2015).
Four of the cohort or longitudinal studies were based on old data from 1987-2004 (Table 1).  (2007), a cohort study is generally prospective and employs an epidemiological approach in the direction of exposure to outcome, or cause to effect. Minimizing bias is important in all cohort studies. A problem in this systematic review was that most of these studies described their design differently and not very clearly (Table 2). In cohort or longitudinal studies, data are collected at two or more time points over an extended period (Polit & Beck, 2006).
The long duration of data collection is a disadvantage due to the cost in terms of time, effort and resources. Another disadvantage is the threat to internal validity including "testing," "mortality" (loss to follow-up) and the influence of confounding variables (Schneider et al., 2007). Social desirability bias can be possible, where informants respond in a way they believe is congruent with the researchers' expectations. Four studies described the small sample size (Tables 1 and 2), which can increase bias. Two studies lacked consideration of how to minimize bias (Agrawal & Keshri, 2014;Burns, Browning, & Kendig, 2015). Four studies were based on self-report questionnaires (Table 1). According to Shadish, Cook, and Campbell (2002), self-reports can increase the likelihood of response bias. Self-reports can be seen as a retrospective data sampling method that increases bias due to the informants' poor memory and inability to remember what actually happened (Polit & Beck, 2012), thus decreasing validity. Self-reports can also be related to unmeasured confounders.
Most of the studies lack references to methodology.
Two studies were described as prospective (Table 1). A prospective design begins with independent variables and looks forward to the effect (Polit & Beck, 2006).
Six of the cohort studies identified important confounding factors in the design and/or analysis (Tables 1 and 2). The follow-up of participants was complete and of adequate duration (Table 2).
No descriptions of validation or the reliability of the measurement instruments used were provided in five of the cohort studies (Table 1). Four of the studies did not refer to the response rate (Table 3). Generalization or external validity was not mentioned in five studies, which can constitute a limitation (Table 2).

| Characteristics of the substance of the quantitative studies
The key aspects that contributed to the evidence from the 12 quantitative studies are presented in Table 3. One of the themes that emerged was: emotional challenges related to experiences of bereavement, depression and anxiety based on the sub-theme social support as the main strategy for coping with emotional pain and suffering. The second theme was: struggling with poor physical health. Four studies found that the participants suffered from depression, anxiety and/or complicated grief disorder (Burns et al., 2015;Carr et al., 2014;DiGiacomo et al.,2013;Ghesquiere et al., 2013;Panagiotopoulos et al., 2013). Those interviewed within a month of their late spouse's birthday reported despair and depression (Carr et al., 2014). In the survey by Ghesquiere et al. (2013), 35% of the widowed participants met the criteria for depression and/or complicated grief disorder on the first measurement occasion, when it was found that 77 met the early criteria for complicated grief disorder, 27 had co-occurring depression, and 12 met the criteria for depression without complicated grief disorder. Of this group, 65 participants were interviewed later and 22 met the criteria for complicated grief disorder, of whom five had co-occurring depression and nine depression without complicated grief disorder. The study by Panagiotopoulos et al. (2013) revealed that Greek-born widows differed from their British-born counterparts, scoring significantly lower on self-rated health and significantly higher on depression and loneliness. Spahni et al. (2015) examined the differences between a widowed and a married sample in relation to the experience of loss.

| Emotional challenges related to experiences of bereavement, depression and anxiety
Three groups were identified and labelled: the resilient group, the vulnerable group and the coper group. The resilient group (N = 215) showed a significantly higher level of extraversion, lower neuroticism and a higher level of consciousness, agreeableness and openness than the copers (N = 155) and the vulnerable group (N = 30).
All differences in psychological resilience between the three profiles were statistically significant. In addition, the resilient group reported a significantly longer time since their loss than the copers. The resilient group and the coper group showed more positive emotional valence concerning the experience of loss than the vulnerable group (Spahni et al., 2015).
One study found that widows reported more depressive symptoms than widowers (Tiedt et al., 2016).
Social support as the main strategy for coping with emotional pain and suffering Eight studies revealed that the older widows and widowers managed their emotional pain and suffering thanks to support described as social bonds, family support or support from friends and that their health problems made daily life a struggle (Agrawal & Keshri, 2014;DiGiacomo et al., 2013;Ghesquiere et al., 2013;Jeon et al., 2013;Perkins et al., 2016;Tiedt et al., 2016;Xu et al., 2017;Zhou & Hearst, 2016). In four studies, the participants needed support from their children and families (Jeon et al., 2013;Panagiotopoulos et al., 2013;Tiedt et al., 2016;Xu et al., 2017), as well as from other people (DiGiacomo et al.,2013;Ghesquiere et al., 2013). Three types of social tie were described, namely contact with statistically significant others, cohabitation with married children and relationships with their children (Jeon et al., 2013). Two studies reported more instrumental support and companionship from sons than from daughters and daughters-in-law (Tiedt et al., 2016;Xu et al., 2017). Transitions to widowhood exhibited a correlation with depressive symptoms, while receiving support was related to a lower level of depressive symptoms (Tiedt et al., 2016;Xu et al., 2017). Instrumental support from daughters correlated with increased depression, while companionship from a daughter-in-law correlated with reduced depressive symptoms (Tiedt et al., 2016). Widowed elders who were living with sons and daughters derived clear benefits compared with those not living with children (Tiedt et al., 2016). The perception of children's filial piety (caring ability) was significantly related to a lower level of depressive symptoms (Xu et al., 2017). Living with sons and daughters was found to reduce depressive symptoms. The results indicated that worry about not having a caregiver was significantly associated with depression. A study from Australia found that the participants perceived a high degree of emotional support from family and friends (Panagiotopoulos et al., 2013). The Greekborn sample perceived significantly greater instrumental support from family, whilst the British-born sample described greater support from friends. For the British-born, loneliness was significantly correlated with family emotional and instrumental support, while emotional support from friends was significantly correlated with loneliness and depression. For the Greek-born, well-being was significantly correlated with family emotional support, with strong correlations between the variables of health, depression and loneliness compared with the British-born sample. Loneliness was also significantly correlated with family instrumental support, while emotional support from friends was reported to a lesser degree compared with the British-born sample. The value of being part of a support group was described in two studies (DiGiacomo et al., 2013;Ghesquiere et al., 2013). Being part of a support group was associated with reduced grief severity, while support from a religious leader was associated with reduced depression severity. However, support from a family doctor was not associated with changes in anxiety, depression or grief severity (Ghesquiere et al., 2013). None of the support types were associated with changes in anxiety severity (Ghesquiere et al., 2013).
Differences in the need for social support were found between western and eastern countries (Panagiotopoulos et al., 2013;Tiedt et al., 2016;Xu et al., 2017). In eastern countries, the participants often lived with their married children and perceived support from them (Jeon et al., 2013;Tiedt et al., 2016;Xu et al., 2017). However, relationships with children only had an impact on depressive symptoms for widowers, whereas cohabitation with children only had an impact on such symptoms for widows (Jeon et al., 2013). Attending support groups decreased significantly with age (Ghesquiere et al., 2013).

| Struggling with poor physical health
Six of the included studies described the informants as having multimorbidity or chronic diseases that caused them to struggle with poor physical health (Agrawal & Keshri, 2014;DiGiacomo et al., 2013;Jeon et al., 2013;Perkins et al., 2016;Tiedt et al., 2016;Zhou & Hearst, 2016) (Table 4). Three of the studies mainly focused on depression but also included information on chronic diseases (DiGiacomo et al., 2013;Jeon et al., 2013;Tiedt et al., 2016). Another three studies mainly focused on chronic diseases but included mental health measures (Agrawal & Keshri, 2014;Perkins et al., 2016;Zhou & Hearst, 2016). Two studies demonstrated that long-term illness had a negative impact on the participants' daily functioning (DiGiacomo et al., 2013;Tiedt et al., 2016) (Table 4). Agrawal and Keshri (2014) revealed that the prevalence of communicable diseases was lower among older widows compared with older widowers. Perkins et al. (2016) found that a widower who had impaired cognitive ability and a mental disorder experienced poorer health. Furthermore, they indicated that both recent and long-term widows were at risk of poorer health compared with married women.
There was no evidence that widowhood acted as a protective factor for either gender.
Age differences were revealed in three studies (Agrawal & Keshri, 2014;Perkins et al., 2016;Zhou & Hearst, 2016). Widowers' physical health was in decline and dropped more rapidly after the age of 70 (Zhou & Hearst, 2016). In addition, two studies revealed that the prevalence of non-communicable diseases increased significantly with age and the same pattern was observed for other types of disease (Agrawal & Keshri, 2014;Perkins et al., 2016).

| D ISCUSS I ON
Twelve studies from different parts of the world are included in this systematic review. Two themes emerged from the thematic analysis (Table 3): emotional challenges related to experiences of bereavement, depression and anxiety based on the sub-theme social support as the main strategy for coping with emotional pain and suffering.
The second theme was: struggling with poor physical health.
The emotional challenges related to the experience of bereavement, depression and anxiety are illustrated by the descriptions of the emotional pain and suffering that constitute the state of widowhood. Between 10%-20% of widowed persons reported that the bereavement had an impact on their quality of life (QOL; Bonanno et al., 2005). Bereavement triggers intense emotions, such as sadness, loneliness, meaninglessness and hopelessness. These emotions influence the health of widows/widowers, leading to a lack of energy, activity and pleasure, culminating in social withdrawal and isolation (Gillies & Neimeyer, 2006). For many years, the emotions following bereavement have been described as universal, exposing a person to a higher risk of mental health problems (Holm, 2009). Recent research by Barrett (2017) revealed that emotions are socially constructed and related to an individual's experiences, context and social relationships. In the early 20th century, Freud suggested that grief might resemble depression, anxiety and posttraumatic stress (Gillies & Neimeyer, 2006). It has been argued that the responses to widowhood constitute grief disorder (Prigerson & Maciejewski, 2005) including intrusive thoughts, memories and images of the loss. Bereavement seems to shatter a widowed person's world, evoking a sense of meaninglessness, unworthiness and overwhelming distress (Gillies & Neimeyer, 2006). Widowed persons attempt to survive by numbing themselves emotionally, but unwanted thoughts and memories still intrude when they try to make sense of their loss. Rebuilding the self can resolve this existential dilemma (Gillies & Neimeyer, 2006). Bereavement and grief share characteristics such as distress and depression that affect health (Jacobsen, Zhank, Block, Maciejewski, & Prigerson, 2010). A century ago Freud stated that self-esteem, self-loathing and suicidality are associated with depression, but not grief (Freud, 1989;Prigerson, Vanderwerker, & Maciejewski, 2008). A previous study has found that narrative intervention can be used effectively for complicated grief disorder (Barbosa, Sám, & Rocha, 2013). Older widowed persons have derived benefit from narrating their stories and working through their loss and grief (Barbosa et al., 2013;Boelen, Keijser, van den Hout, & van den Bout, 2007). Grief has many symptoms in common with depression (Latham & Prigerson, 2004;Simon et al., 2005).
Prolonged grief disorder is distinguished by the severity and duration of symptoms, as well as the marked distress and disability they evoke (Prigerson et al., 2008 Social support is often described as the help provided by one's social network (Haber, Cohen, Lucas, & Baltes, 2007). Structural support is given by persons in the social network and related to the frequency of contact within the network. Functional support includes emotional and instrumental support as well as perceptions of and judgements about the support. There is a cultural difference between eastern and western countries, despite the fact that an individual's emotional and pain system is described as universal and developed from pain mechanisms that originated millions of years ago. Barrett (2017) views emotions as social constructions. The emotional pain system has been explained by Panksepp and Watt (2011) as social cohesion, which forges bonds between infants and caretakers, fortifies friendships and sexual relationships and promotes social solidarity among groups of living species. Arousal of this system can be related to social attachments, thus explaining how much one misses someone and why depression hurts so much (Panksepp & Watt, 2011). Explanation of bereavement and depression seems to be rooted in the attachment theory (Bowlby, 1980), which can facilitate understanding that bereavement involves emotions and that it is not essential to distinguish depression from prolonged grief disorder. Similar to this review study, Guiaux, Van Tilburg, and Van Groenou (2007) found an increase in emotional and instrumental support during the first 2 years of widowhood that decreased over time.
Filial piety is one of the guiding principles of elder care in Chinese culture, emphasizing respect, loyalty and support for older parents (Mjelde-Mossey, Chi, & Lou, 2005). Even in Japan, obligations to support elderly parents seem to be mainly rooted in traditional Confucian ideals, where the eldest son has financial responsibility for his parents' home and property, while his wife is expected to assume caretaking responsibilities (Therborn, 2004). However, developments in recent decades have led scholars to believe that filial piety may be less pervasive in contemporary Japan (Traphagan, 2003). Powers, Bisconti, and Bergeman (2014) revealed that structural and functional support remained stable across the first 2 years loss. However, emotional and social support seemed to decrease over time.
Struggling with poor physical health is supported by previous research. Studies identified increased mortality from causes such as heart diseases, cirrhosis of the liver, accidents and suicide (Stroebe, Schut, & Stroebe, 2007). As shown in this review, poor physical health in widowhood is most apparent in eastern countries (Moon, Glymour, Vable, Liu, & Subramanian, 2014;Shor et al., 2012). Previous research reported that poor health and/or loneliness in widowhood can increase morbidity and mortality (van den Berg, Lindeboom, & Portrait, 2011;Williams, 2005;Wittstein et al., 2005). Poor mental health can result in suicide (Möller et al., 2011). van den Berg et al. (2011) found strong effects of widowhood on mortality because of the increase in multimorbidity. Interventions soon after bereavement are vital for the length and QOL in widowhood. Mortality has been referred to as "the widowhood effect" and different mortality-related causes have been reported (Möller et al., 2011;Stroebe et al., 2007). The widowhood effect seems to increase the need for physical and mental health care to reduce the risk of death both in eastern and western countries.
From a stress perspective, it is argued that the combined effects of spousal death, cardiovascular disease and/or depression could make widowed older adults vulnerable to early death (Stahl et al., 2016). Research on gender revealed that older widows are more likely than older widowers to suffer from multimorbidity (Agrawal & Keshri, 2014). Widows are more prone to poor health and financial insecurity (Williams, Baker, Allman, & Roseman, 2006). This is supported by three of the included studies (Perkins et al., 2016;Xu et al., 2017;Zhou & Hearst, 2016), especially in the eastern part of the world. Poor health has been associated with age, education and socio-economic conditions (Agrawal & Keshri, 2014).

| Methodological discussion of the included studies
This systematic review has limitations due to the quality of the included studies (Table 2), the fact that many have only a small sample (Table 1) and lack information about methodological characteristics (Table 1). In addition, no studies from the Nordic countries or from Africa were found. One must also take into consideration that different cultures could have a severe impact on elderly widows and widowers. Psychological well-being seems to be neglected in countries such as Pakistan, India, China and South Korea. These countries appear to focus more on the curative rather than the preventive aspects of health care (Khan, 2012). Such differences can be a limitation when reviewing studies from different parts of the world.
Another methodological limitation that could increase bias involves emotions and the fact that the roles of widowers and widows are dynamic and socially constructed (Barrett, 2017;DiGiacomo et al., 2013;Williams et al., 2006). Gender differences are more apparent in societies with rigid gender roles such as South Korea, India, China and even Japan, as outlined in four of the included studies (Jeon et al., 2013;Perkins et al., 2016;Xu et al., 2017;Zhou & Hearst, 2016). In eastern countries, the economic burden of widowhood is far more apparent than in western countries. Poor socio-economic status can be related to increased use of healthcare services (Agrawal & Arokiasamy, 2009;Halleröd & Gustafsson, 2011). In high-income countries, a better health status has been reported for widows than for widowers. Such findings seem to be inconsistent with studies from low-income countries, where marriage has been described as a benefit for men. Thus, widowhood seems to be influenced to a great extent by culture and ethnicity (Kleinmann, 2012;Zhou & Hearst, 2016).
Traditional gender roles in these countries have placed widows in a subordinate position to widowers and although they persist in the older South Korean and Chinese population, they seem to be weakening (Jeon et al., 2013). Culture, gender and age appear to play an important role in the experience of loss (Jeon et al., 2013;Xu et al., 2017;Zhou & Hearst, 2016). Stereotypes of older people need to be considered because of the differences in what is acceptable behaviour for widows and widowers in western and eastern countries.
An important point to bear in mind is that such differences make it difficult to draw conclusions about the health burden in countries not included in this review.

| CON CLUS ION
Healthcare professionals need more knowledge and skills to address the factors that influence the health of older widowed adults. Future research should explore the lived experiences of the health problems and burden of widowhood. Healthcare professionals must take action to minimize the risk of mortality due to health problems and multimorbidity in this population. It is important to build a community team that can contact widows and widowers in the first year of bereavement to map their unmet need for help, thus preventing prolonged grief, physical health problems and mortality.

| CONTRIBUTI ON S
ALH: Study design, ALH: data collection, and ALH, AB and ES: data analysis, discussion and preparation.

ACK N OWLED G EM ENTS
The authors wish to thank Monique Federsel for reviewing the English language.

E TH I C A L A PPROVA L
As this is a systematic review of previous studies, a patient consent statement and Research Ethics Committee approval were not required.

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