The Effect of Economic Conditions at Birth, Education, and Social Integration on Dementia

(1)

The Effect of Economic Conditions at Birth,

Education, and Social Integration on

Dementia

C. K. Kiernan Joint Master Thesis

June 2020

Keywords: Dementia; early economic conditions; business cycle; duration analysis

Word count: 6,491

Supervisor: Professor Dr. G. van den Berg

Student number: s3434966

(2)

Abstract

(3)

1. Introduction

Recent decades have been characterized by demographic changes, resulting from increased life expectancy and lower fertility rates (Christensen et al., 2009). This has led to a relatively higher number of elderly populations. Ageing populations have significant consequences for social systems and the economy (Reher, 2011). These consequences are two-fold: older individuals offer less labour, and thus contribute less taxes to fund the healthcare system, while also requiring more healthcare (Bloom et al., 2010). While improved living conditions and medical care have led to healthier elderly (Doblhammer et al., 2013b), it may also give rise to diseases which were not as prevalent in previous older generations.

One such disease is Major Neurocognitive Disorder, commonly known as Dementia. Dementia is a term for several diseases characterized by progressive loss of memory and other cognitive abilities, which affect a person’s ability to maintain daily living (Hugo and Ganguli, 2014; Livingston et al., 2017). The loss of cognitive ability and memory can have significant impact on an individual’s life. Due to diminished independence, those diagnosed with dementia require extensive care. Families and caregivers may be impacted by the long-term nature and high demands of dementia care, causing many caregivers to experience negative physical and mental health outcomes themselves (Romero-Moreno et al., 2012; Hugo & Ganguli, 2014). In terms of resources, Leicht et al. (2011) estimated the annual net costs of dementia to be €15 000 for mild dementia cases, €32 000 for those experiencing moderate dementia, and €40 000 for those in the severe category. It is evident that dementia does not only impact the individual, but also their social network, the healthcare system, and the economy (Doblhammer et al., 2013b; Hugo & Ganguli, 2014). These significant implications and consequences of dementia are why prevention and delaying onset are important areas of study.

(5)

cardiovascular disease, heart disease, obesity and depression (Livingston et al., 2017). Studies suggest that there may be modifiable factors, which can act protectively against dementia (Livingston et al., 2017). The Cognitive Reserve Hypothesis suggests that those with more cognitive reserve may develop dementia later than those with less (Livingston et al., 2017; Stern, 2002). The theory suggests that an individual’s cognitive reserve may be enhanced through physical and cognitive stimulation (Livingston et al., 2017). Building on the Cognitive Reserve Hypothesis, this study will focus on the effect of education and social integration on dementia onset. As such, the research question of this study will be:

Are higher levels of education and social integration associated with delayed onset of dementia?

I will study this question using duration modelling. For this purpose, I will make use of data from the first eleven waves of the LISS (Longitudinal Internet Studies for the Social sciences) panel administered by CentERdata (Tilburg University, the Netherlands). Following van den Berg et al. (2006), I will focus on macroeconomic conditions as an indicator of economic conditions around birth and early childhood, as these are exogenous at the individual level and thus circumvent simultaneity bias. For this purpose, macroeconomic data were retrieved from the Maddison Project Database 2018.

(6)

2. Theoretical Framework

2.1. Dementia

Dementia is a term for a group of disorders characterized by a decline in cognitive level and progressive loss of memory (Livingston et al., 2017). Common forms of dementia include Alzheimer’s disease, vascular dementia, dementia with Lewy bodies, and mixed dementia (Livingston et al., 2017). In the Netherlands, it is common for the General Practitioner (GP) to diagnose dementia. If necessary, they will refer the patient to a specialist. According to the Dutch College of General Practitioners (NHG) guidelines, GPs ought to take a three-step approach, which includes signalling symptoms, diagnostics and performing an inventory of care requirements. Common signs of dementia include memory problems, socially inappropriate behaviour and limitations in functioning (Dieleman-Bij de Vaate et al., 2020). For those experiencing early-onset dementia, generally before the age of 65, changes in behaviour, personality and functioning may be more apparent than memory problems (Dieleman-Bij de Vaate et al., 2020). A qualitative study by Prins et al. (2016) indicated that GPs often only pursue diagnosis if the cognitive limitations cause the patient to experience problems in functioning. Furthermore, they found that a specific diagnosis was important for younger patients, while the need for specialist referral was found to be less for older ones, as younger patients may need to make more preparations for future care (Prins et al., 2016). Daily living and social functioning of those diagnosed with dementia is greatly affected, with many no longer able to live independently and requiring extensive long-term care (Hugo & Ganguli, 2014; Livingston et al., 2017).

2.2. Age and General Health Status

(7)

certain threshold after which the accumulated damage manifests itself in the form of severe cognitive impairment or dementia (Fratiglioni and Wang, 2007). Dementia prevention should therefore aim to avoid or delay this threshold (Fratiglioni and Wang, 2007; Livingston et al., 2017). It has been suggested that there are modifiable risk factors during the life course, which may be targeted to reduce dementia risk. These include early childhood conditions and the cognitive resilience (Livingston et al., 2017).

2.3. Early Life Conditions

Thrifty Phenotype Hypothesis. The thrifty phenotype hypothesis was first introduced by Hales and Barker (2001). They related deprivation in utero and early childhood to development of type 2 diabetes later in life (Hales and Barker, 2001). This hypothesis has been extended to other areas, stating that poor fetal and infant nutrition can lead to increased risk of chronic disease later in life. The underlying mechanisms can be constructed as follows: adverse conditions, such as low income, can lead to reduced access to nutrition, quality housing, and medical care, which can obstruct the ability of a child’s brain to develop to its fullest potential (Moceri et al. 2001). Studies have shown that early life economic conditions can be associated with various later life health outcomes, such as mortality (van den Berg et al., 2006), mental health (Angelini et al., 2019), and cardiovascular disease (Alessie et al., 2019). The effect of early life economic conditions has also been found in cognitive ability (Doblhammer et al., 2013b; Fritze, 2014) and Alzheimer’s disease (Moceri et al., 2001). Therefore, the following hypothesis is formulated:

Hypothesis 1: Poor economic conditions at birth will be associated with earlier dementia

onset.

There are various ways to construct early life economic conditions, including considering individual conditions and macroeconomic conditions.

(8)

mental health, measured childhood socioeconomic status through the number of rooms, facilities, and books in the home, and occupation of the main breadwinner.

Macroeconomic Conditions. Several studies suggest studying macroeconomic conditions in early life instead of individual conditions, as macroeconomic conditions are exogenous at the individual level and thus circumvent the simultaneity bias resulting from unobserved heterogeneity affecting both conditions around birth and later life health outcomes (van den Berg et al., 2006). Macroeconomic conditions can be studied in various ways, for instance through fluctuations in unemployment rates (Alessie et al., 2019) or through the business cycle at the time of birth, measured in terms of Gross National Product (van den Berg et., 2006) or Gross Domestic Product (Doblhammer & Fritze, 2015). The underlying mechanism has been described in studies by Doblhammer et al. (2013a) and Fritze et al. (2014). Individuals who are currently in the dementia risk category in terms of age, were born during times with lack of a social safety net (Doblhammer & Fritze, 2015; Doblhammer et al., 2013a). A recession would then lead to severe adverse conditions, such as low quality and quantity of nutrition, psychological stress, poor housing conditions and limited access to health (Doblhammer & Fritze, 2015; Doblhammer et al., 2013a). A study by Doblhammer et al. (2013a) of 17,070 individuals in ten European countries found that birth during a recession had significant effects on cognitive abilities in later life. Similar results were found in Fritze et al. (2014).

2.4. Cognitive Reserve

The Cognitive Reserve Hypothesis is a theory which can function as a potential mechanism for coping with the deficits that lead to dementia (Livingston et al., 2017; Stern, 2002). More cognitive reserve is thought to improve the ability to cope with more brain damage, while those with more cognitive reserve may also be able to use brain networks more efficiently (Scarmeas and Stern, 2003; Stern, 2002). In this way, more cognitive reserve may delay the threshold of dementia manifestation (Stern, 2002). Factors contributing to the cognitive reserve include educational and occupational attainment, leisure activities and social integration (Scarmeas et al., 2001; Stern, 2002; Scarmeas & Stern, 2003).

(9)

Stern, 2003). Scarmeas et al. (2001) performed a longitudinal community-based cohort study of individuals aged above 65 and found 38% less risk of developing dementia in those with high leisure activity. These activities included reading, walking, excursions, going to movies or restaurants, and visiting with friends and family (Scarmeas et al., 2001). A study by Verghese et al. (2003) found that activities such as reading, playing board games, playing musical instruments, and dancing were associated with reduced dementia risk, even after individuals suspected of preclinical dementia were excluded.

Education. Education may also contribute to the cognitive reserve. According to Stern (2002), individuals with higher levels of educational attainment, having acquired more cognitive reserve, can process tasks more efficiently, and thus sustain greater brain damage before showing signs of dementia (Stern, 2002). This may be a direct effect of education. Additionally, higher educational attainment can lead to higher income jobs and safer work environments (Cutler and Lleras-Muney, 2006), which, as an indirect effect, can improve access to health and health insurance. Therefore, educational attainment may lead to improved general health. Fritze et al. (2014) identified education as a possible moderator of living conditions in childhood on later life cognitive abilities. They found that low education is associated with greater risk of cognitive impairment. Russ et al. (2013) found that low educational attainment was associated with increased risk of dementia death in women but found no relationship for men. Takasugi et al. (2019) found that one year of education is associated with a 17% decrease in the risk of Alzheimer’s disease. Therefore, the following hypothesis is formulated:

Hypothesis 2: Education will be associated with delayed dementia onset.

(10)

Hypothesis 3: Social integration will be associated with delayed dementia onset. Figure 1 depicts the conceptual model of this study and highlights some possible pathways between early childhood conditions, education, social integration, and dementia. As discussed above, I will hypothesize that poor early childhood conditions will increase the risk of dementia onset, while education, by improving cognitive abilities directly and health indirectly, will reduce risk of dementia onset. Furthermore, I hypothesize that social integration, by delaying cognitive decline and improving general health, will reduce risk of dementia onset.

Figure 1: Conceptual model. The solid lines indicate direct effects, while the dashed lines indicate

(11)

3. Data

Previous studies have identified economic conditions at birth, education and social integration as possible predictors of dementia onset. While leisure activities have also been identified as a possible predictor, it will not be considered in this study because of small sample size. Two separate datasets were combined in order to conduct the analysis. For individual level data, the LISS (Longitudinal Internet Studies for the Social sciences) panel administered by CentERdata (Tilburg University, The Netherlands)1_{was used. The Maddison Project Database 2018}2_was used to acquire historic macroeconomic data. First, the individual level data collection and variable construction will be described, followed by the macroeconomic data collection and variable construction. An overview of the variables considered in this study and their definitions can be found in Appendix B.

3.1. Individual Level Data

Data Collection. For individual level data on dementia, education, and social integration, I made use of the data from the LISS panel. The LISS panel is a representative sample of Dutch individuals. The panel is based on a true probability sample of households drawn from the population register. If necessary, households are provided with a computer and Internet connection. Individuals participate in monthly Internet surveys and a longitudinal survey is fielded in the panel every year, covering a large variety of domains3_{. The Background Variables} survey was used to acquire data on year of birth, gender, education, and domestic situation. The Health survey was used to acquire dementia event data.

Study Sample. For each survey, waves 1 to 11 were used, covering an eleven-year time period ranging from 2007 to 2018. There were instances of panel attrition, which leads to an unbalanced panel. These respondents were not dropped, as this would significantly reduce sample size. Respondents with missing birth year or MND variables were dropped from the sample. Four respondents were dropped from the sample, as their year of birth changed between waves. All individuals born after 1970 were dropped from the sample, as they may be too young to be considered part of the dementia risk group. In the remaining sample, no individuals under

1_{More information about the LISS panel can be found at: www.lissdata.nl} 2_{More information about the Maddison Project Database 2018 can be found at:}

https://www.rug.nl/ggdc/historicaldevelopment/maddison/releases/maddison-project-database-2018

(12)

the age of 40 or above the age of 90 indicated MND. Therefore, those below and above these ages were not considered in the analysis. The Health survey was not held in the year 2014. In the following year, there were two cases of MND. It is assumed that those individuals received MND diagnosis in the year 2014 and their survival time was adjusted to reflect this. The final sample of this study consisted of 7,480 subjects.

Major Neurocognitive Disorder. The outcome variable in this study is dementia. The Health survey was used to acquire dementia event data. In the survey, respondents were asked whether a physician had diagnosed them with Alzheimer, dementia, organic brain syndrome, senility, or another serious memory problem, in the last year. As there is no way to distinguish which condition the respondent suffers from, this study considered them as one combined variable, named Major Neurocognitive Disorder (MND).

Education. The Background Variables survey was used to acquire data on education. In the survey, education is measured in terms of the respondent’s highest level of education in Statistics Netherlands (CBS) categories. For the purpose of this study, the CBS categories were aggregated into the following categories: primary education if the respondent’s highest level of education was primary school; secondary education if the respondent’s highest level of education was VMBO or HAVO/VWO; vocational education if the respondent’s highest level of education was MBO; and higher education if the respondent’s highest level of education was HBO or WO. Each of these variables is a dummy variable, equal to 1 if it is the highest level of education completed, and zero otherwise.

Social Integration. In order to test the hypothesis concerning social integration, I considered the domestic living situation of the respondent. The Background Variables survey was used to acquire this variable. In the survey, respondents can select which of the following categories applies to their situation: single; (un)married cohabitation, without children; (un)married cohabitation, with children; single with children; or other. For the purpose of this study, these categories are aggregated into the following categories: single; (un)married cohabitation; and

(13)

3.2. Macroeconomic Data

Data Collection. Part of the aim of this study is to compare individuals who were born during challenging economic times to those who were born in more fortunate times. I used the business cycle at birth as an indicator of economic circumstances at and around birth (Doblhammer and Fritze, 2015). For this purpose, GDP data from the Maddison Project Database 2018 is used. This database provides historic real GDP per capita data for a variety of countries, expressed in 2011 US dollars (see Bolt et al., 2018 for more information on this project). I considered data from the Netherlands for the time period 1910 – 1980.

Business Cycle at Birth. Following van den Berg et al. (2006), I decomposed the annual logarithm of GDP per capita into a trend and cyclical component using a Hodrick-Prescott (HP) filter. Several smoothing parameters were applied in order to find one which provided appropriate smoothing. While studies suggest a smoothing parameter of 6.25 for annual data (Ravn and Uhlig, 2002), I applied a HP-filter with smoothing parameter 2500 as it provided better smoothing of the trend. An overview of the considered smoothing parameters can be found in Appendix A.

(14)

The macroeconomic data will be used to construct two variables which measure economic conditions at birth. The cyclical component of the HP-filter is used to construct the variable

birth recession. This is a dummy variable which indicates whether the individual was born

during a recession, that is, when the logarithm of GDP is below its trend value. In order to address the circumstances during the Hunger Winter, a dummy variable is introduced to indicate whether the respondent was born in 1944 or 1945. This variable will be included in the basic duration model in order to examine whether birth during the Hunger Winter drives the results.

Figure 2: log annual GDP per capita and trend component

Figure 3: cyclical component

8 8.5 9 9.5 10 1910 1920 1930 1940 1950 1960 1970 1980 year

(15)

4. Research Methods

In order to study the impact of the identified predictors on dementia onset, I will conduct two types of analysis. First, a logit regression will be conducted, followed by a duration analysis. All analyses were conducted using STATA SE version 16.1.

4.1. Logit Model

First, I conducted a logit regression. I sought to examine how the predictor variables birth recession, age, female, education and domestic situation affect MND diagnosis. I only considered those who reported MND in their first observation, i.e. the first wave in which they first participated. The outcome variable is MND, which equals 1 if the respondent has indicated MND diagnosis in their first observation, and 0 otherwise. Note that this analysis considers individuals who have indicated MND diagnosis in later waves as not having MND. The estimated model includes background variables, such as age and gender, education and domestic situation.

The logit model can be specified as:

Pr{𝑀𝑁𝐷 = 1|𝑥!}

= 𝐿(𝛽_"+ 𝛽_#𝑏𝑖𝑟𝑡ℎ𝑟𝑒𝑐𝑒𝑠𝑠𝑖𝑜𝑛 + 𝛽_$𝑎𝑔𝑒 + 𝛽_%𝑓𝑒𝑚𝑎𝑙𝑒 + 𝛽_&𝑠𝑒𝑐𝑜𝑛𝑑𝑎𝑟𝑦 + 𝛽'𝑣𝑜𝑐𝑎𝑡𝑖𝑜𝑛𝑎𝑙 + 𝛽(ℎ𝑖𝑔ℎ𝑒𝑟 + 𝛽)𝑐𝑜ℎ𝑎𝑏 + 𝛽*𝑜𝑡ℎ𝑒𝑟 + 𝜀!)

where 𝐿 =_#,++ . MND is a dummy variable indicating whether individual 𝑖 indicated MND diagnosis. The result of this analysis is depicted in table 2 and will be further discussed in the section 5.

4.2. Duration Model

(16)

since the year of birth. As the 2015 survey was held in August rather than December, some individuals have the same age during two different observations. Since this is not possible in the duration analysis, I use the constructed variable survival years instead of the age reported by the respondent. The onset of risk is considered to start at birth, and respondents enter the analysis at the age of first observation. Respondents exit the analysis when they indicate MND diagnosis or when they are censored.

Censoring is a common phenomenon in duration analysis. The data considered in this study involves right censoring, as individuals may have developed MND after the 11th_{wave, having} their diagnosis remain unobserved. Right censoring also occurs if the individual drops out of the study without having been diagnosed with MND. Left censoring may occur if the individual already has MND before entering the study. In this sample, thirteen individuals indicated MND diagnosis in the first year in which they were observed. These individuals were not considered in this analysis, as it could not be determined when they were diagnosed.

In order to study the hypotheses, I conducted duration analysis using Cox proportional-hazards models. This model is used as it does not require distributional assumptions (George et al., 2014). I examined four models. First, I considered a basic model, examining the relationship between business cycle and MND diagnosis, controlling for gender. Next, I extended this model to include birth during Hunger Winter. The second model introduced education and the third model introduced social integration. Finally, I examined a full model considering all variables.

The full model can be specified as:

ℎ(𝑡, 𝑥) = ℎ_"(𝑡) exp(𝛽#𝑏𝑖𝑟𝑡ℎ𝑟𝑒𝑐𝑒𝑠𝑠𝑖𝑜𝑛 + 𝛽$𝑓𝑒𝑚𝑎𝑙𝑒 + 𝛽-ℎ𝑢𝑛𝑔𝑒𝑟𝑤𝑖𝑛𝑡𝑒𝑟

+ 𝛽_%𝑠𝑒𝑐𝑜𝑛𝑑𝑎𝑟𝑦 +𝛽_&𝑣𝑜𝑐𝑎𝑡𝑖𝑜𝑛𝑎𝑙 + 𝛽_'ℎ𝑖𝑔ℎ𝑒𝑟 + 𝛽₍𝑐𝑜ℎ𝑎𝑏 + 𝛽₎𝑜𝑡ℎ𝑒𝑟)

(17)

5. Results

5.1.Logit Model

The first part of the analysis consists of a logit regression, examining those who reported MND in their first observation. Table 1 provides characteristics of the respondents in the sample during the first survey in which they participated. The total sample consists of 7,480 subjects. Out of the 43 total MND cases, 13 occurred during the first observation. It can be seen that nearly 70% of MND cases were born in favourable times. This may already be an indicator that the hypothesis regarding business cycle at birth does not hold. About 31% of MND cases were female and 38.5% of MND diagnoses occurred between the ages of 50 and 59. None of the MND cases were born during the Hunger Winter. Vocational education was the highest level of education for approximately 46% of MND cases and 84.6% described their domestic situation as (un)married cohabitation.

Table 1: Respondent Characteristics (baseline)

Variable No MND (n = 7,467) MND (n = 13) Total (n =7,480) Gender 0 (Male) 1 (Female) 3,507 3,960 47.0% 53.0% 9 4 69.2% 30.8% 3,516 3,964 47.0% 53.0% Age Mean 40 – 49 50 – 59 60 – 69 70 – 79 80 – 89 56.7 2,276 2,243 1,914 847 187 30.5% 30.0% 25.6% 11.3% 2.5% 55.8 4 5 2 1 1 30.8% 38.5% 15.4% 7.7% 7.7% 56.7 2,280 2,248 1,916 848 188 30.5% 30.1% 25.7% 11.3% 2.5%

Business cycle at birth

< 0 ≥ 0 2,8854,582 38.6% 61.4% 4 9 30.8% 69.2% 2,889 4,591 38.6% 61.2% Hunger Winter 0 1 7,103 364 95.1% 4.9% 13 0 100% 7,116 364 95.1% 4.9% Primary education Secondary education Vocational education Higher education Missing 746 2,892 1,486 2,329 24 10.0% 38.7% 19.9% 31.2% 0.3% 1 3 6 3 7.7% 23.1% 46.2% 23.1% 747 2,895 1,492 2,332 24 9.5% 38.1% 20.5% 31.6% 0.3% Domestic Single (un)married cohabitation Other 1,951 5,466 50 26.1% 73.2% 0.7% 1 11 1 7.7% 84.6% 7.7% 1,952 5,477 51 26.1% 73.2% 0.7%

(18)

The results of the logit regression can be found in Table 2 and are reported as odds ratios. It is evident that there is a lack of statistically significant result for the effect of birth during a recession on MND diagnosis. This holds for most of the predictor variables. Only the domestic situation other is statistically significant, implying that the odds of MND diagnosis increases times 41.5 compared to single. I now proceed to conduct a duration analysis to study all the respondents who indicated MND diagnosis.

Table 2: Logit model

(19)

5.2.Duration Model

In this analysis, I observed 6,027 subjects, of which 30 experienced MND. Respondents who dropped out after one wave or were left-censored were excluded from analysis. Table 3 depicts the results of the Cox proportional hazards analysis. It can be seen that none of the models examining the effects of business cycle on MND onset yield any statistically significant results. In model 4, it can be observed that only (un)married cohabitation becomes statistically significant at the 10% level. This may suggest that those who cohabitate experience earlier onset of dementia, compared to individuals who live alone. This does not support the hypothesis that social integration is associated with delayed dementia onset.

(20)

It can be argued that the lack of statistically significant result in the previous analyses is due to the small number of MND cases in the dataset. In order to examine whether this is a possibility, a duration model is used, in which all individuals who drop out of the study are assumed to have MND. Those who remain until the last wave in 2018 will be considered censored. As in the previous analysis, respondents who drop out after one wave or are left-censored are excluded from analysis. The results for this duration model can be found in Table 4. With the increased number of individuals with MND diagnosis, several coefficients have become statistically significant. As these results are based on the assumption that all individuals who drop out of the study do so because they have MND, I cannot derive conclusions from them for the purpose of this study. However, it does indicate that larger sample sizes are necessary in order to study dementia onset.

In Model 1, no statistically significant results are found for business cycle at birth. The variable

female is significant, and this might indicate that being a woman is associated with slightly

earlier onset of dementia. The same is evident when the model is extended to include the Hunger Winter. The inclusion of Hunger Winter does not appear to affect the variable female, but it is evident that the sign of the business cycle changes.

In Model 2, female is still associated with earlier onset of dementia than in men. Education is significant for each variable. Secondary, vocational and higher education are each associated with delayed onset of dementia, compared to primary education.

In Model 3, the model is extended to include social integration. Female is still associated with earlier onset of dementia than in men, and the domestic situation variable other is associated with a delayed onset of dementia, compared to single.

(21)

Table 4: Cox proportional hazards model – all dropouts considered MND Covariates Model 1 Basic Model 2 Education Model 3 Social Model 4 Full No Hunger Winter Hunger Winter Female 0.095 (0.038)∗∗ 0.095 (0.038)∗∗ 0.074 (0.038)∗ 0.094 (0.038)∗∗ 0.073 (0.038)∗ Birth Recession 0.0015 (0.043) −0.0018 (0.044) −0.0036 (0.044) −0.003 (0.044) −0.004 (0.044) Hunger Winter 0.034 (0.101) 0.044 (0.101) 0.034 (0.101) 0.0453 (0.101) Education Primary Secondary Vocational Higher Reference −0.123 (0.065)∗ −0.297 (0.074)∗∗∗ −0.229 (0.068)∗∗∗ Reference −0.123 (0.065)∗ −0.296 (0.074)∗∗∗ −0.230 (0.068)∗∗∗ Domestic Single (un)married cohabitation Other Reference −0.042 (0.044) −0.345 (0.146)∗∗ Reference −0.034 (0.044) −0.348 (0.146)∗∗ Sample Failures 𝑁 = 6,022 2,802 𝑁 = 6,006 2,791 𝑁 = 6,022 2,802 𝑁 = 6,006 2,791 𝒑∗∗∗_{≤ 𝟎. 𝟎𝟏; 𝒑}∗∗_{≤ 𝟎. 𝟎𝟓; 𝒑}∗_{≤ 𝟎. 𝟏}

(22)

6. Discussion and Conclusion

In this paper, I examined the effects of economic conditions around birth, education and social integration on dementia onset. The LISS panel was used to acquire individual level data. GDP data for the Netherlands from the Maddison Project Database 2018 was used to construct a macroeconomic indicator for conditions around birth. Cox proportional hazards models were employed to conduct duration analysis.

6.1. Discussion

Due to the lack of statistical significance, I cannot conclude that the trends evident in the results are valid. After extending the duration model to assume that all respondents who drop out of the study prematurely were diagnosed with MND, significant results were found for women,

secondary, vocational and higher education, and for domestic situations which can be described

as other. A similar study should be executed on larger data samples.

The lack of significant results may be stemming from the limited number of MND cases in the sample. The incidence rate of MND in this study was only 43, of which 13 indicated it in their first observation, against a total sample of 7,480. This is much lower than numbers reported in Francke et al. (2018), who estimate prevalence of dementia in the Netherlands to be around 254,000 – 270,000 in 2016, against a population of nearly 17 million. However, as indicated in the report by Francke et al. (2018), these figures are based on numbers accumulated from multiple studies, some of which were executed over 25 years ago and in a very specific context (a suburb of Rotterdam), so the validity of these numbers may be disputed. Regardless, in order to effectively study the effect of business cycle at birth, education and social integration on dementia, more data on dementia in the Netherlands is necessary.

(23)

communicate effectively. This also raises questions of consent and privacy. Individuals with dementia may not be able to provide consent to participate in a survey. Another issue may be that individuals with dementia require extensive care and may move to full-time care facilities. Individuals in such facilities may not be targeted for surveys.

In the case of this study, there may also be some concerns about the reliability of the data. As it is self-reported, there is no way to verify the accuracy of the answers supplied by respondents. This can have several implications. First, individuals who were diagnosed with dementia may not report it for various reasons. Second, answers provided in other surveys by those who have been diagnosed with dementia may not be accurate. While they have their own limitations, physician-reported or insurance data may be more accurate than self-reported survey data, and therefore a better alternative in analyses of dementia onset.

Other studies have found effect of economic conditions around time of birth on cognitive abilities late in life. A reason why this study does not attain similar results may in part be due to the fact that dementia has a stricter definition, compared to cognitive decline. Cognitive functioning has been measured through numeracy, verbal fluency and recall ability (Doblhammer et al., 2013a; Fritze et al., 2014). On the other hand, the definition of dementia in the context of this study requires a formal diagnosis by a physician or specialist. According to guidelines by the Dutch College of General Practitioners(NHG), dementia diagnosis can be made if there is a significant cognitive decline which interferes with daily functioning, and which cannot be explained by other medical conditions (Dieleman-Bij de Vaate et al., 2020). This definition is thus stricter than that of cognitive decline.

6.2. Limitations

(24)

other health-related conditions. Biological, genetic and health-related factors were not accounted for in this study. Not including these in analysis may be considered a limitation.

6.3. Further Research and Conclusion

Due to the limited sample size in this study, I was unable to study the effects of leisure activities. For future research, it would be interesting to do an analysis of the effect of various leisure activities on dementia onset. Additionally, for future research it would be interesting to study the impact of the Internet. Since the late 1990s, people have started to rely more heavily on the Internet and other forms of technology to communicate or conduct leisure activities, such as online gaming. Individuals may participate less in organized activities or face-to-face social contact than they did before. This raises questions about ways to measure social integration and leisure activities. Future research could include those types of interaction and activities when examining the effects on dementia onset.

(25)

References

Angelini, V., Howdon, D., & Mierau, J. (2019). Childhood socioeconomic status and late- adulthood mental health: Results from the survey on health, ageing and retirement in europe. The Journals of Gerontology. Series B: Psychological Sciences and Social

Sciences, 74(1), 95-104.

Angelini, V., & Mierau, J. O. (2014). Born at the right time? Childhood health and the business cycle. Social Science and Medicine, 109, 35–43.

Alessie, R., Angelini, V., Van den Berg, G., Mierau, J., & Viluma, L. (2019). Economic conditions at birth and cardiovascular disease risk in adulthood: Evidence from post- 1950 cohorts. Social Science and Medicine, 224, 77-84

Bloom, D.E., Canning, D. and Fink, G. (2010). Implications of population ageing for economic growth. Oxford Review of Economic Policy, 26(4), 583-612.

Christensen, K., Doblhammer, G., Rau, R., & Vaupel, J. (2009). Ageing populations: The challenges ahead. The Lancet, 374(9696), 1196-1208.

Cutler, D.M. and Lleras-Muney, A., (2006). Education and Health: evaluating theories and evidence. National Bureau of Economic Research Working Paper Series. 12352 Dieleman-Bij de Vaate AJM, Eizenga WH, Lunter-Driever PGM, Moll van Charante EP,

Perry M, Schep-Akkerman A, Smit BSJ, Starmans R, Verlaan-Snieders MNE, Van der Weele GM (2020). NHG-Standaard Dementie (versie 5). Huisarts & Wetenschap. 63(4)

Doblhammer, G., van den Berg, G.J., Fritze, T., (2013a). Economic Conditions at the Time of Birth and Cognitive Abilities Late in Life: Evidence from Ten European Countries.

PLoS ONE 8(9)

Doblhammer, G., Fink, A., Fritze, T., & Gunster, C. (2013b). The demography and epidemiology of dementia. Geriatric Mental Health Care, 1(2), 29-33.

Doblhammer, G and Fritze, T. (2015). Month of Birth and Dementia Late in Life. KZfSS

Kölner Zeitschrift für Soziologie und Sozialpsychologie. v67 nS1: 217-240

Francke AL, van der Heide I, de Bruin S, Gijsen R, Poos R, Verbeek M, Wiegers, T, Willemse B. (2018). Een samenhangend beeld van dementie en dementiezorg: kerncijfers, behoeften, aanbod en impact. Themarapportage van de Staat van Volksgezondheid en Zorg.

Fratiglioni, L. and Wang, H. (2007). Brain Reserve in Dementia. Journal of Alzheimer’s

Disease, 12: 11-22

(26)

George, B., Seals, S., & Aban, I. (2014). Survival analysis and regression models. Journal of

Nuclear Cardiology, 21(4), 686–694

Giles, L.C., Anstey, K.J., Walker, R.B. and Luszcz, M.A. (2012). Social networks and memory over 15 years of follow up in a cohort of older Australians: Results from the Australian Longitudinal Study of Ageing. Journal of Aging Research.

Hales, C.N, & Barker, D.J.P. (2001). The thrifty phenotype hypothesis. British Medical

Bulletin, 60(1), 5-20

Hugo, J., & Ganguli, M. (2014). Dementia and cognitive impairment: Epidemiology, diagnosis, and treatment. Clinics in Geriatric Medicine, 30(3), 421-442

Leicht, H., Heinrich, S., Heider, D., Bachmann, C., Bickel, H., van den Bussche, H., … König H.-H. (2011). Net costs of dementia by disease stage. Acta Psychiatrica

Scandinavica, 124(5), 384–395.

Livingston, G. et al (2017). Dementia prevention, intervention, and care. The Lancet, 390(10113), 2673-2734)

Maddison Project Database, version 2018. Bolt, Jutta, Robert Inklaar, Herman de Jong and Jan Luiten van Zanden (2018), “Rebasing ‘Maddison’: new income comparisons and the shape of long-run economic development

Moceri, VM., Kukull, WA., Emanual I., et al. (2001). Using census data and birth certificates to reconstruct the early-life socioeconomic environment and the relation to the

development of Alzheimer's disease, Epidemiology, 12, pp. 383-389

Painter, R. C., Roseboom, T. J., & Bleker, O. P. (2005). Prenatal exposure to the dutch famine and disease in later life: an overview. Reproductive Toxicology, 20(3), 345–352

Prins, A., Hemke, F., Pols, J. & Moll van Charante, E.P. (2016) Diagnosing dementia in Dutch general practice: a qualitative study of GPs’ practices and views. British

Journal of General Practice. 66(647); 416-422.

Ravn, M. O., & Uhlig, H. (2002). On adjustings the Hodrick-Prescott filters for the frequency of observations. Review of Economic Studies, 69, 371–375

Reher, D.S. (2011). Economic and social implications of the demographic transition.

Population and Development Review. 37, 11-33.

Romero-Moreno, R., Losada, A., Mrquez-Gonzlez, M., & Mausbach, B. T. (2012). Variables modulating depression in dementia caregivers: a longitudinal study. International

Psychogeriatrics, 24(8), 1316–1324

Russ, T.C., Stamatakis, E., Hamer, M., Starr, J.M., Kivimäki, M., and David Batty, G. (2013). Socioeconomic status as a risk factor for dementia death: individual participant meta- analysis of 86 508 men and women from the UK. The British Journal of Psychiatry. 203, 10-17.

(27)

Scarmeas, N., Levy, G., Tang, M., Manly, J., & Stern, Y. (2001). Influence of leisure activity on the incidence of Alzheimer's disease. Neurology, 57(12), 2236-42

Scarmeas, N., and Stern, Y. (2003). Cognitive Reserve and Lifestyle. Journal of Clinical and

Experimental Neuropsychology. 25(5), 625-633.

Sharp, E.S. and Gatz, M. (2011). Relationship between education and dementia; an updated systematic review. Alzheimer disease and associated disorders. 25(4): 289-304 Stern, Y. (2002). What is cognitive reserve? Theory and research application of the reserve

concept. Journal of International Neuropsychological Society. 8, 448-460.

Song, X., Mitnitski, A., & Rockwood, K. (2014). Age-related deficit accumulation and the risk of late-life dementia. Alzheimer's Research & Therapy, 6(5-8)

Sörman, D.E., Rönnlund, M., Sundström, A., Adolfsson, R., and Nilsson, L. (2015). Social relationships and risk of dementia: a population-based study. International

Psychogeriatrics. 27:8, 1391-1399.

Takasugi, T., Tsuji, T., Nagamine, Y., Miyaguni, Y., & Kondo, K. (2019). Socio‐economic status and dementia onset among older Japanese: A 6‐year prospective cohort study from the Japan Gerontological Evaluation Study. International Journal of Geriatric

Psychiatry, 34(11), 1642–1650.

Tay, L., Tan, K., Diener, E., and Gonzalez, E. (2013). Social relations, health behaviors, and health outcomes: a survey and synthesis. Applied Psychology: Health and Well-Being. 5(1), 28-78.

Van den Berg, G., Lindeboom, M., & Portrait, F. (2006). Economic conditions early in life and individual mortality. American Economic Review, 96(1), 290-302.

Verghese, J., Lipton, R., Katz, M., Hall, C., Derby, C., Kuslansky, G., . . . Buschke, H. (2003). Leisure activities and the risk of dementia in the elderly. The New England

Journal of Medicine, 348(25), 2508-16

(28)

Appendix A: HP-Filter Smoothing Parameters

The figures below show the graphs of the logarithm of annual per capita GDP in the Netherlands, using various smoothing parameters. Smoothing parameters of 6.25, 100, 500 and 1000 were applied to find an appropriate smoothing parameter. When applying these filters, it can be seen that trend fluctuations too much.

Figure 4: Smoothing Parameter 6.25

Figure 5: Smoothing Parameter 100

8 8.5 9 9.5 10 1910 1920 1930 1940 1950 1960 1970 1980 year

log annual GDP per capita trend component (HP-Filter) smoothing parameter: 6.25 8 8.5 9 9.5 10 1910 1920 1930 1940 1950 1960 1970 1980 year

(29)

8 8.5 9 9.5 10 1910 1920 1930 1940 1950 1960 1970 1980 year

log annual GDP per capita trend component (HP-Filter) smoothing parameter: 500 8 8.5 9 9.5 10 1910 1920 1930 1940 1950 1960 1970 1980 year

(30)

Appendix B: Variable Definitions

Variable Description Value

Background Age Gender

Age of the respondent.

Indicator of gender of the respondent.

years 1 if female, 0 if male Macroeconomic Birth Recession Hunger Winter

Measures business cycle at birth.

Indicator of whether respondent was born during Dutch Famine.

1 if business cycle is below 0, and 0 otherwise, 1 if born in 1944 or 1945, 0 otherwise Education Primary Secondary Vocational Higher

Indicators of highest level of

education obtained. 1 if yes, 0 if no 1 if yes, 0 if no 1 if yes, 0 if no 1 if yes, 0 if no Social Integration Single Cohab Other

Indicates whether the respondent’s domestic situation can be described as single.

Indicator of whether the respondent’s domestic situation can be described as (un)married cohabitation.

Indicator of whether the respondent’s domestic situation can be described as other than single or (un)married cohabitation.

1 if yes, 0 if no

The Effect of Economic Conditions at Birth, Education, and Social Integration on Dementia