Elderly retirement postponement factors: an integrative literature review

This study aimed to search for scientific evidence published in analytical studies which address elderly workers’ retirement postponement factors. An integrative literature review was performed searching for papers in Virtual Health Library (BVS) and SciELO databases from January 2009 to March 2018. Twelve papers were selected following consensus by independent reviewers, generating two categories: National Literature and International Literature. We chose this subject because the phenomenon of population aging entails different social and economic impacts between developed and developing nations, such as Brazil. In the analysis of the results, we found that elderly retirement decision involves a variety of personal, organizational and other factors (legislative, cultural, socio-political, technological). In conclusion, we note scarcity of studies on the subject, with exclusive samples of older adults in both categories, as well as that the greater or lesser relevance of a given factor will depend on the life context of each worker. Moreover, since retirement decision is a dynamic process, there is a shortage of longitudinal studies that need to be met with more robust national surveys of the elderly population.


introduction
Global population aging has been occurring rapidly as a result of the fall in fertility rate in several regions and higher life expectancy 1,2 . In developing countries, the phenomenon of population aging has taken place in a relatively short time, and social gains have not adequately accounted for the majority of the population 3,4 .
In Brazil, the pace of population aging is relevant. With an estimate of the United Nations population projection, the proportion of the elderly population in the country should increase from 11.7% to 23.5% within 24.3 years from 2015 2 .
The greater longevity, that is, the increase in the mean number of years to be lived by the population should be considered for planning the social security system, which includes social assistance, health and social security 2,4 . Currently, extending the working life of older workers is a priority on the agenda of many Western countries that seek to increase retirement ages in line with increased longevity 5 and the social need of workers to extend their working lives 6,7 . However, the actual labor market exit age is determined by the structural factor of this market and the social security system, as well as by the economic conditions, preferences and individual conditions 8 .
Moreover, it is essential to consider the age factor as the one that stands out the most in the labor relationships, especially concerning the elderly workers 9 . Age is also an important variable to explain activity rates, so a consequence of population aging is the aging of the economically active population (EAP). Thus, since proportionally fewer young people enter the labor market, workers will have to remain longer in the economic activity to offset this unbalance. This extended working life may be necessary for the elderly, not least because of the income generated, but also because of the prominence of work in human sociability 10 .
Concerning the Western world, work is one of the substrates of self-esteem, personal identity and the individual's sense of utility 11 . Therefore, in a predominantly capitalist society, work finds itself in a prioritized space in people's lives. Moreover, in this society that values only those who produce, a retired older adult can show negative feelings for the loss of the professional role 12 .
Regarding the dependency ratio, this is an indicator to measure the relative participation of the potentially inactive population segment, which should be sustained by the portion of the potentially productive population. Thus, the age group of the population defined as economically dependent/inactive comprises individuals younger than 15 years and those aged 60 and over, while the potentially productive/active segment includes those aged between 15 and 59 years 2 .
It is relevant to comment that this indicator may not necessarily reflect the economic dependency ratio, due to circumstantial factors that affect the labor market, not considering several situations. Thus, the most important ones are the incorporation of young people, the permanence of the elderly or the exclusion of people of productive age 10 .
According to data of the Tenth section of the Situation Charter of April 2018 available on the website of the Institute of Applied Economic Research, Brazilian older adults are the least representative portion of the total employed population. However, this proportion has increased, up from 6.3% to 7.8% in the period from 2012 to 2018 13 .
This may be a consequence of population aging and a possible change in the behavioral pattern of older workers about their decisions regarding retirement, which must be ratified in the future. The increase of these older workers is probably due to a possible change in the behavior of Brazilians in this age group over their decisions to remain in the labor market 13 .
In Brazil, older adults have played an increasing role in the composition of household income. Information from the Brazilian Institute of Geography and Statistics (IBGE) of 2013 showed that 65% of this population was the primary source of income of their households and that 15.6% of elderly retirees and pensioners worked to supplement income 4,14 .
Thus, we must consider the various factors that can intervene in the work-retirement decision, which entails three possible paths: definitive retirement, postponement of retirement with work in the same job, or work after retirement 15 . Therefore, retirement does not always represent a break with the world of work 7,12 . Older employees are increasingly accepting bridge employment, which occurs when older workers take paid employment after retirement from their primary career 7,16 .
According to the literature, the determinants of postponement of retirement among the older adults may involve personal aspects 6,17,18 , work-related aspects 7,8 and other external aspects that encompass the socio-political context, legis-lation, culture and new technologies 19,20 . Finally, although the worker has the option to choose whether or not to retire, the decision is conditioned by the limits and opportunities the world provides 21 .
Understanding why and knowing the factors that keep older workers in organizations extending their working lives will help managers and society understand older people better and formulate public policies in favor of this age group. Thus, the higher interest must be towards making professional life more sustainable for people 22 . Also, it is vital that work continuity could be thought from the perspective of well-being provided by the very nature of work and the environment in which these people find themselves 12 .
On the other hand, while political decisions in many countries plan to raise the official retirement age in the coming decades, little research has been conducted on the factors that influence the intention of older employees to continue working after the normal retirement age 23 . Considering the increased longevity, studies focused on older adults and their participation in the world of work are relevant. These data will help clarify the theme by increasing the knowledge about this recent phenomenon, which is extending the working life with older adults remaining in the institutions, as well as encourage other research, stir reflections and originate actions geared to the well-being and improved quality of life of Brazilian elderly workers, improving their work context.
Given these considerations, this study was carried out through an integrative review of the literature on this subject. Thus, this study aimed to analyze the published scientific evidence that portrays the factors related to the postponement of retirement by elderly workers.

Methods
This is a descriptive study of integrative review of literature, whose method facilitates the search, critical analysis and synthesis of the knowledge produced on the subject investigated and may include studies of several methodologies. Thus, it allows the identification of knowledge gaps, guidance for future research and the incorporation of the applicability of significant results in practice [24][25][26] .
The following steps were followedto accomplish this integrative review: 1) Identification of the theme and elaboration of the guiding ques-tion of the research; 2) Establishment of the inclusion and exclusion criteria and search in the database; 3) Categorization of selected studies; 4) Evaluation of studies included in the review; 5) Analysis and interpretation of results; and 6) Presentation of the review and synthesis of knowledge 23 .
The following guiding question was elaborated from the theme: What are the scientific pieces of evidence published in the last nine years that portray the factors related to References were searched with MeSH (Medical Subject Headings) and DeSC (Health Sciences Descriptors) controlled descriptors intermediated by the Boolean operators "AND" and "OR", and, thus, the search strategies occurred with the following key in the advanced search: Retirement OR Jubilación OR Aposentadoria AND Aged OR Anciano OR Idoso OR "Aged, 80 and over" OR "Anciano de 80 o más anos" OR "Idoso de 80 anos ou mais". We opted for this broad key because the searches on other keys were not fruitful when we included uncontrolled descriptors such as "Retirement age"; "Retirement decision"; "Older workers"; Postpone; Postponement; Postponing; Procrastinate, among others.
The following inclusion criteria for selection of references were established: original paper whose full text is available online and with content about factors that lead the elderly to postpone retirement; published in a scientific journal in English, Spanish and Portuguese in the period 01/01/2009-22/04/2018.
It is relevant to consider that due to the low availability of papers with an exclusive sample of older adults, this integrative review considered works mixed with seniors and older or mature workers (40 years and over). Review papers, theses, dissertations, books, editorials, reviews and government documents were excluded.
Initially, papers were selected by two independent reviewers, who read titles, abstracts and keywords carefully to verify adequacy to inclu-sion and exclusion criteria. After removing duplicate papers, a new screening was performed with full-text reading, eliminating those that did not include the defined objectives and selecting the papers for review after consensus among reviewers.
In the third step of the review, we used an instrument recommended and validated by Ursi 27 to extract the data of the selected papers, which enabledthe synthesis of each one, preserving their differences 26 .
We then proceeded to evaluate the methodological rigor through an instrument adapted from the Critical Appraisal Skills Program (CASP) of the Public Health Resource Unit (PHRU), prepared by the University of Oxford 27 . This adapted tool had 10 items (maximum 10 points), and ranking works with the following scores: 6-10 points, good methodological quality and reduced bias; and minimum 5 points, satisfactory methodological quality with increased risk of bias (Table 1).
In the studies, an instrument of the US Agency for Healthcare and Research and Quality (AHRQ) was also applied. In this review, we followed Stillwell et al. 28 , who rank hierarchically by seven levels of evidence according to the research design (Table 2).
Next, the papers were categorized and analyzed, followed by the synthesis and interpretations of the results. Finally, the synthesis of the knowledge with a discussion of the results and proposals for future studies was performed.

results
Initially, the search in the databases found 13,795 productions, of which -after filtering using the criteria of inclusion of language, years and type of document -11,338 papers were eliminated. After reading the titles, abstracts and keywords of the remaining 2,457 works and exclusion of duplicates, 53 papers were left for full-text reading, of which 41 did not meet the eligibility criteria, leaving 12 in the final selection that underpinned this review (Figure 1).
The twelve papers were submitted to the instrument recommended by Ursi 27 to gather and synthesize the critical information, and following categorization of the papers. All the papers had a score ≥ 7 according to a critical analysis using CASP, showing the good quality of the methodology employed with reduced bias 29 .
The analysis of the studies allowed the construction of two categories: national literature and international literature. Categorization is justified because the former has several similar characteristics such as language; they were conducted in a developing country with vast social inequalities, and where people aged 60 and over are considered older adults. On the other hand, the international papers found (all in English) derive from developed countries, most of them with high Human Development Indexes (HDI), and where older adults are those who aged 65 years and over. Tables 3 and 4 show the synthesis of the characterization and other relevant information of the selected studies by categorization of national and international literature, respectively.

National literature
All papers in this category were classified as level VI of evidence because they showed evidence derived from a single descriptive or qualitative study 28 .
Research by Menezes and França 30 investigated variables in the retirement decision in public servants aged between 45 and 67 years. Through multiple logistic regression analyses, these authors found that age, job perception, working hours' flexibility, and job control predict the decision to postpone retirement. It is worth mentioning two findings. First, only 25% of the participants chose to work in retirement for income supplementation, and second, the variable high life expectancy was not significant.
Pires et al. 17 conducted a study on senior nursing professionals with the possibility of retirement, whose results showed that the financial factor was the leading influencer to stay in the job. Other factors found by these authors were: appreciation of the work environment and own activities; work was a strategy against idleness and residential space; positive perception regarding work, and negative perception about retirement.
Authors Figueira et al. 18 showed that in a pre-retired group, in the age range of 48-60 years, family and work experiences influence retirement decision-making, and depending on the context, may lead to its postponement or not.
The study by Macêdo et al. 15 was carried out with 283 public servants, aged between 48 and 69 years, who could retire or were less than five years from retirement. The results revealed that the decision to postpone retirement occurred to employees who perceived themselves as working, with no replacement activities, and a non-attractive domestic environment.

international literature
All the papers in this category were located in MEDLINE, five of which were classified as ev-  Nilsson et al. 31 conducted a study in Sweden with health sector employees age 55-64 years, and among the conclusions emphasized that mental and physical work environment, work rhythm, skill and competence were essential factors in determining whether older workers believed they could work up to 65 years of age or older.
In Finland, Perkiö-Mäkelä and Hirvonen 32 in a large sample of employees aged 45-63 years showed that they considered continuing to work when there was a low physical workload associated with one or more of the following factors: good health, financial gains, significant and challenging work, and flexible workload.
Authors Oakamana and Howie 19 conducted a study on focus groups of 42 senior civil servants (over 40) in Australia and found that flexible working conditions and environments and excit-   33 research with four-year follow-up with participants in Denmark's early retirement scheme was concentrated in the 60-64 age bracket that could retire. They concluded that high job satisfaction and positive psychosocial working conditions could cause older employees to postpone retirement.
Wind et al. 7 found in a sample of Dutch retirees aged 57-67 that work motivation, physical health, household financial situation and volunteering played an essential role in working beyond retirement.
Hofäcker and Naumann 34 analyzed data from three available rounds (years 1996, 2002 and 2011) of the German Aging Survey (DEAS), which is a national cross-sectional longitudinal study of the population aged 40 and over. The results showed that the change in the political paradigm of the German early retirement modified the employment behavior of older workers, with an increased working time. These authors suggest that the probability of the most educated and less educated workers to work beyond the retirement age converge in quantitative terms; however, the reasons differ because the former wish voluntarily while the latter are motivated by financial necessity. The conclusion of the study reveals that the motivating factor of choosing to continue working beyond retirement can reinforce social inequalities.
Cho et al. 8 conducted an eight-year longitudinal study to identify the determinants of the retirement process in wage earners over 50 years of age using a comparison of data from South Korea, Germany and US. Statistics showed that more than 40% of employees worked after retirement in the three countries.
The empirical analysis of this Korean study showed that the receipt of smaller pensions leads to postponement of retirement or work after retirement. Germany has a well-funded social security system that allows older people with low  To evaluate factors within 9 areas that influence why some older workers want or can work up to 65 or more Perkiö-Mäkelä and Hirvonen, Finland, 2012 (32) Physical workload and thoughts of retirement Work Quantitative Cross-sectional n = 2,013 Interviews To examine the opinions of employees about continuing work until the official retirement age and after the 63 years of age, and find out if there is any relationship with the workload Oakmana and Howie, Australia, 2013 (19) How can organisations influence their older employees' decision of when to retire? Work Qualitative Interviews with Focal Group n = 42 To report a study on older employees to a large public service organization and examine their intentions to retire.
Hofäcker and Naumann, Germany, 2015 (34) How can organizations influence their older employees' decision of when to retire? Work To classify the retirement process and empirically identify the individual and institutional characteristics that determine the retirement process of the elderly schooling to be more likely to retire. In contrast, Korean older adults receive negligible retirement pensions that they cannot get out of paid work, even if they wished so. In the US, no statistical-ly significant impact was found between public pensions and the interaction of occupation with an education level. This result can be explained because Americans believe that public pensions are more targeted to low-income classes, while corporate pensions prevail for most workers 8 .

Discussion
In the search for references in the databases, among the 12 papers selected, only the one carried out by Pires et al. 17 in Brazil comprised only elderly workers. Thus, there was a shortage of scientific publications in the Brazilian and global setting on factors of postponement of the retirement bysenior workers with a sample of only senior individuals by the WHO chronological criteria 35 .
Regarding the research methods used, the papers of this review evidenced a predominance of qualitative methods in the Brazilian literature and quantitative methods in the international scope. In the latter, we emphasize sets of research with robust samples and longitudinal studies that strengthen them. This finding corroborates the reports on the lack of national studies using longitudinal research methods 36 . On the other hand, as a strength for national research, the authors Hofäcker and Naumann 34 report that quantitative research possibly reaches its analytical limits and that complex thematic such as a retirement decision are benefited and captured by qualitative approaches better 4 .
Similar to what is already consolidated in the world literature, most of the papers selected evidenced an agreement with the variable health or subjective perception of health status as a predictor for the final retirement. In the study carried out in Finland, the results showed that health itself was stated as a reason to consider continuous work beyond the official retirement age, especially by employees with a high physical workload 32 . However, Swedish work reports that although people need good health and good ability to work, the economy seemed to be more critical than health status among study respondents 31 .
According to Stafford et al. 37 , education and occupational status may confuse or modify the association between physical or cognitive ability and retirement. These authors report that less educated workers or lower-status jobs may be less able to make adaptations at work, while tasks involving higher-status jobs tend to require a higher educational level and may require a higher level of cognitive ability 37 . This report is compatible with the findings of this integrative review. Indeed, in the results of the study by Wind et al. 7 , the authors identified that the educational level could only partially explain which factors predicted to work beyond retirement.
Moreover, the German study considered that the reasons for working beyond retirement differed by education, and this was considered one of the main determinants of social inequality 34 . The authors of this research further claim that the emerging trend of work beyond retirement reinforces social inequalities, since lower-educated workers may not be able to meet the demands or even benefit from policies that encourage prolonged working life. That is, in contrast to highly skilled workers, the retirement decision of low-educated workers is mainly driven by economic needs and is often not compatible with the level of employment or the ability to work longer 34 .
The Korean study examined the determinants of the retirement process by comparing three developed countries with different national labor markets and social institutions. The empirical analysis in this research showed that a small public pension leads to a later retirement, and suggests that the level of schooling exerts positive or negative correlation according to the level of welfare and financing of the national social security 8 .
The results of the research by Menezes and França 30 did not confirm the findings of the authors Solinge and Henkes 5 that the subjective life expectancy influences the intention of the workers' retirement decision since this variable was not statistically significant. However, one strength of the Dutch work with a longitudinal design was the possibility of examining retirement intentions together with the actual retirement behavior. These authors suggest that some older workers may have little control over the time of retirement or retirement is not always voluntary.
An interesting finding to be highlighted in the national study by Menezes and França 30 was that only 25% of the participants stated that they would continue to work for economic reasons, showing that the option to work in retirement is not always stimulated by the need to supplement incomes 30 .
About life partners and close friends, both national papers and those of other countries showed a consistent performance of these factors concerning whether older workers want to postpone retirement and have a positive or negative correlation as per the experience of each.
Analyzing the conclusions of the researches of the two categories, we can observe that they are in agreement with regard to the vari-ous work-related factors that affect the decision of older workers to continue working beyond the retirement age, such as motivation at work, commitment to career, job perception, flexible working hours, job requirements, job control, satisfaction with their functions and valuation, development possibilities, positive management relationships, good community and enjoyable working environment, without age discrimination, among others.In the Danish study, it is relevant to consider that high job satisfaction was the strongest psychosocial predictor of late retirement 33 . The Australian study recommends a regular review of employee performance and satisfaction to maximize productivity and retention of older workers 19 .

Final considerations
Despite being a worldwide phenomenon, population aging brings about different social and economic impacts between developed and developing nations, such as Brazil. Age transition is occurring rapidly in terms of national status, without the population having reached the social achievements and real improvement of living conditions 3 . Therefore, one questions the generalization of the international results to the Brazilian population, considering that HDI levels, social security rules, social, political and economic conditions are different from other countries 36 . Furthermore, the cross-sectional nature of national surveys did not facilitate the dynamic analysis of the determinants of postponement of retirement since participants were only asked about the process. Also, follow-up studies were not performed to verify whether the intentions mentioned will turn into real behavior at the time of retirement 23 . We can conclude that the lack of cohort studies in this area is a gap that must be bridged with more robust national studies.
Based on the considerations made, further studies are required on the expectations of Brazilian senior workers regarding retirement, a better understanding of the aging process, the factors that determine the participation of professional life in old age and that affect the decisions about the extended professional life.
Moreover, the strategies of retaining the workforce in the profession must be combined with improved working conditions, which is a determining factor for the permanence of these workers 38 . If rulers encourage older workers to postpone retirement, on the other hand, they must provide the necessary support to those who wish to retire 15 . However, according to Haddad 39 , although the aging population is increasing worldwide, the economic model exerts power over all borders and governments and does not make investments to improve social protection systems for older adults 39 . collaborations LBC Souza: conception and design of the integrative review; selection, analysis and categorization of articles in the integrative review; analysis of data and interpretation of results, writing of the manuscript and approval of the version of the article to be published. MCC Leal: study design; data analysis and interpretation of results; writing and revision of the manuscript; and approval of the version of the article to be published. AFB Bezerra: revision of the manuscript and approval of the version of the article to be published. ICL Silva: independent reviewer 1 with categorization, analysis and selection of articles from the integrative review; and writing and editing tables and figures. LCC Souza: independent reviewer 2 with categorization, analysis and selection of articles from the integrative review; and writing and editing tables and figures. ACG Espírito Santo: final approval of the version to be published.