ASSOCIATIONS OF SUICIDE RATES AND AFFECTIVE DISORDERS WITH LITHIUM LEVELS IN DRINKING WATER

LITHUANIAN UNIVERSITY OF HEALTH SCIENCES MEDICAL ACADEMY

Vilma Liaugaudaitė

ASSOCIATIONS OF SUICIDE RATES

AND AFFECTIVE DISORDERS

WITH LITHIUM LEVELS

IN DRINKING WATER

Doctoral Dissertation Medical and Health Sciences,

Dissertation has been prepared at the Neuroscience Institute (until 2017.12.31 Behavioral Medicine Institute) of Medical Academy of Lithuanian University of Health Sciences during the period 2016–2021.

Scientific Supervisor

Prof. Dr. Nida Žemaitienė (Lithuanian University of Health Sciences, Medical Academy, Medical and Health Sciences, Public Health – M 004).

Consultant

Prof. Dr. Narseta Mickuvienė (Lithuanian University of Health Sciences, Medical Academy, Medical and Health Sciences, Nursing – M 005).

Dissertation is defended at the Public Health Research Council of the Lithuanian University of Health Sciences:

Chairperson

Prof. Dr. Ričardas Radišauskas (Lithuanian University of Health Sciences, Medical Academy, Medical and Health Sciences, Public Health – M 004).

Members:

Prof. Dr. Žemyna Milašauskienė (Lithuanian University of Health Sciences, Medical Academy, Medical and Health Sciences, Public Health – M 004);

Assoc. Prof. Dr. Rugilė Ivanauskienė (Lithuanian University of Health Sciences, Medical Academy, Medical and Health Sciences, Public Health – M 004);

Prof. Dr. Valentina Vengelienė (Vilnius University, Natural Sciences, Biology – N 010);

Prof. Dr. Gil Zalsman (Tel Aviv University, Medical and Health Scien-ces, Medicine – M 001).

Dissertation will be defended at the open session of the Public Health Research Council on the 9th _{of April, 2021 at 12 a. m. in the auditorium} A-202 of the Centre for the Advanced Pharmaceutical and Health

LIETUVOS SVEIKATOS MOKSLŲ UNIVERSITETAS MEDICINOS AKADEMIJA

Vilma Liaugaudaitė

SAVIŽUDYBIŲ IR AFEKTINIŲ

SUTRIKIMŲ SĄSAJOS SU

LIČIO KONCENTRACIJA

GERIAMAJAME VANDENYJE

Disertacija rengta 2016–2021 m. Lietuvos sveikatos mokslų universiteto Medicinos akademijos Neuromokslų institute (iki 2017-12-31 Elgesio medicinos institutas).

Mokslinė vadovė

prof. dr. Nida Žemaitienė (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, medicinos ir sveikatos mokslai, visuomenės sveikata – M 004).

Konsultantė

prof. dr. Narseta Mickuvienė (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, medicinos ir sveikatos mokslai, slauga – M 005).

Disertacija ginama Lietuvos sveikatos mokslų universiteto Visuomenės sveikatos mokslo krypties taryboje:

Pirmininkas

prof. dr. Ričardas Radišauskas (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, medicinos ir sveikatos mokslai, visuomenės sveikata – M 004).

Nariai:

prof. dr. Žemyna Milašauskienė (Lietuvos sveikatos mokslų universi-tetas, Medicinos akademija, medicinos ir sveikatos mokslai, visuomenės sveikata – M 004);

doc. dr. Rugilė Ivanauskienė (Lietuvos sveikatos mokslų universitetas, Medicinos akademija, medicinos ir sveikatos mokslai, visuomenės sveikata – M 004);

prof. dr. Valentina Vengelienė (Vilniaus universitetas, gamtos mokslai, biologija – N 010);

prof. dr. Gil Zalsman (Tel Avivo universitetas, medicinos ir sveikatos mokslai, medicina – M 001).

Disertacija ginama viešame Visuomenės sveikatos mokslo krypties tary-bos posėdyje 2021 m. balandžio 9 d. 12 val. Lietuvos sveikatos mokslų uni-versiteto Naujausių farmacijos ir sveikatos technologijų centro A-202

audi-CONTENT

LIST OF ABBREVIATIONS ... 7

DEFINITIONS OF TERMS ... 8

INTRODUCTION ... 10

AIM AND OBJECTIVES ... 12

1. REVIEW OF LITERATURE ... 14

1.1. Epidemiology of suicide ... 14

1.1.1. Suicide Worldwide ... 14

1.1.2. Suicide in Lithuania ... 15

1.1.3. Suicide prevention ... 15

1.1.4. Suicide and mental and behavioral disorders ... 17

1.2. The Anti-suicidal effect of lithium ... 19

1.2.1. Dietary source of lithium ... 19

1.2.2. The Effects of lithium deficiency on behavioral parameters ... 20

1.2.3. Mechanistic considerations of lithium ... 20

1.3. Summary of studies investigating the anti-suicidal effects of lithium in drinking water ... 21

1.3.1. Analysis of selected studies ... 21

1.3.2. Studies on potential anti-suicidal effects of lithium as a trace element in drinking water ... 26

1.3.3. Summary of the literature review ... 27

2. MATERIAL AND METHODS ... 29

2.1. Study design ... 29

2.2. Methods ... 30

2.2.1. Study area: evaluation of lithium levels in drinking water ... 30

2.2.2. Dependent variable: suicide standardized mortality rate ... 31

2.2.3. Independent variables ... 31

2.3. Statistical analysis ... 32

3.4. Relationships between suicide SMRs, incidence of affective

disorders, and lithium levels ... 48

3.4.1. Analysis of suicide SMRs and other indicators of municipalities according to quartiles of lithium levels... 48

3.4.2. Analysis of suicide SMRs and other indicators of municipalities according to a median of lithium levels ... 49

3.4.3. Associations of suicide SMRs, and mental and behavioral disorders with lithium levels in low lithium exposure group ... 53

3.4.4. Relationships between suicide SMRs, mental and behavioral disorders, and lithium levels in high lithium exposure group ... 55

3.4.5. Exploratory factor analysis of the variables in high lithium exposure group: supplementary analysis ... 60

4. DISCUSSION ... 63

4.1. The lithium level in drinking water of the central wellfields in Lithuanian municipalities ... 63

4.2. The association between suicide SMR and lithium levels in drinking water and across municipalities ... 64

4.2.1. Association between suicide SMR and lithium levels in drinking water: Study I (Pilot study) ... 64

4.2.2. Association between suicide SMR and lithium levels in drinking water: Study II ... 65

4.3. The impact of incidence of affective disorders on an association of suicide SMR with lithium level in drinking water ... 68

LIMITATIONS AND STRENGTH OF THE STUDY ... 75

CONCLUSIONS ... 77

PRACTICAL AND SCIENTIFIC RECOMMENDATIONS ... 78

SANTRAUKA ... 80

REFERENCES ... 106

PUBLICATIONS ON THE DISSERTATION THEME ... 117

ANNEXES ... 137 Annexe 1 ... 138 Annexe 2 ... 139 Annexe 3 ... 140 CURRICULUM VITAE ... 141 ACKNOWLEDGEMENTS ... 143

LIST OF ABBREVIATIONS

AUD – Alcohol use disorder BD – Bipolar disorder CI – Confidential interval

DALYs – Disability Adjusted Life Years DDD – Defined Daily Dose

EU – European Union IQR – Interquartile range

Li – Lithium

MBD – Mental and behavioral disorders MDD – Major depressive disorder

OECD – Organization for Economic Cooperation and Development OR – Odds ratio

PC – Principal components

PCA – Principal component analysis RDA – Recommended Dietary Allowances RR – Relative risk

SA – Suicidal attempt SB – Suicidal behavior SD – Standard deviation

SGA – Second generation antipsychotics SI – Suicidal ideation

SMR – Standardized mortality rates

SPSS – Statistical Package for the Social Sciences USEPA – United States Environmental Protection Agency VIF – Variance Inflation Factor

WHO – World Health Organization WLS – Weighted Least-Squares

DEFINITIONS OF TERMS

Affective disorder – any mental disorder, as depressive disorder, bipolar

disorder in which a major disturbance of feelings or emotions is predominant.

Anxiety disorder – chronic feelings of overwhelming anxiety and fear,

unattached to any obvious source, that can grow progressively worse if not treated. The anxiety is often accompanied by physical symptoms such as sweating, cardiac disturbances, diarrhea or dizziness.

Bipolar disorder – a serious illness that causes shifts in a person’s mood,

energy and ability to function. Dramatic mood swings can move from “high” feelings of extreme euphoria or irritability to depression, sometimes with periods of normal moods in between.

Comorbidity – in general, the existence of two or more illnesses,

whether physical or mental, at the same time in a single individual.

Depression – mental state characterized by feelings of sadness,

lone-liness, despair, low self-esteem, and self-reproach; accompanying signs include psychomotor retardation or, at times, agitation, withdrawal from interpersonal contact, and vegetative symptoms, such as insomnia and anore-xia. The term refers to a mood that is so characterized or to a mood disorder.

Mental disorder – psychiatric illness or disease whose manifestations

are primarily characterized by behavioural or psychological impairment or function, measured in terms of deviation from some normative concept; associated with distress or disease, not just an expected response to a particular event or limited to relations between a person and society.

Lithium – lithium carbonate or other preparations of lithium metal used

to treat manic depression and bipolar disorders.

Psychosis – mental disorder in which the thoughts, affective response,

ability to recognize reality, and ability to communicate and relate to others are sufficiently impaired to interfere grossly with the capacity to deal with reality; the classic characteristics of psychosis are impaired reality testing, hallucinations, delusions, and illusions.

Schizophrenia – a mental illness that is characterized by disturbances in

thought (such as delusions), perception (such as hallucinations), and behavior (such as disorganized speech or catatonic behavior), by a loss of emotional responsiveness and extreme apathy, and by noticeable deterioration in the level of functioning in everyday life.

Substance use disorder – any mental or behavioral disorder resulting

Suicide – a self-inflicted death with evidence (either explicit or implicit)

of intent to die.

Suicide attempt – is defined as a self-inflicted, potentially injurious

behavior with a nonfatal outcome for which there is evidence (either explicit or implicit) of intent to die. A suicide attempt may result in no injuries, injuries, or death.

INTRODUCTION

Suicide is a serious global public health problem that demands our attention but preventing suicide is no easy task [1]. Suicide is the second leading cause of death among 15–29 year olds globally and one of the most important indicators of the public’s state of mental health [2].

Suicide is a complex phenomenon with many contributing factors including psychological, social, economic, biological, and environmental [3– 6]. However, mental health disorders are also attributed to a significant number of indirect deaths through suicide and self-harm. Suicide deaths are strongly linked, although not always attributed to mental health disorders. There is strong evidence that suicide mostly occurs among people with affective disorders [7]. The suicide risk has been estimated at 6–10% in the affective disorder population, which is 10 times the corresponding risk in non-psychiatric populations [8].

Treatment of patients with suicidal behavior is one of the most challenging tasks for health-care professionals [9]. To date, several strategies have been proposed for suicide prevention, both at population and individual level, some of which may be pharmacological [10, 11]. Lithium is the “gold standard” mood stabilizer against which potential mood stabilizer agents are judged [12]. A meta-analysis demonstrated an overall significant efficacy of lithium in preventing suicide with a highly significant reduction in rates of suicide [13–16].

In clinical practice it is well established that lithium has a mood-stabi-lizing and suicide-preventive effect in individuals suffering from affective disorders [8, 15]. A systematic review and meta-analysis of randomized trials on this topic suggested that lithium treatment reduces mortality and suicide by more than 60% in people with major depression or bipolar disorder [17]. A placebo-controlled trial data showed that low doses of lithium might improve and stabilize mood quite rapidly in former drug users [18].

The duration of lithium pharmacotherapy needed for suicide prevention is important. Findings from the international multi-centre trial are compatible with the idea that long-term lithium treatment extends the survival of patients suffering from affective disorders to match the general population. The randomized as well as open clinical trials revealed that the risks of completed and attempted suicide were consistently lower, by approximately 80%, during treatment of bipolar and other major affective disorder patients with lithium for an average of 18 months [19].

growing evidence that even very low lithium levels induced by routine consumption of lithium from tap water may have anti-suicidal effects both in patients with affective disorders, and in the general population [22, 23]. One hypothesis the explaining anti-suicidal effects of low lithium levels is that long-term exposure to lithium through routinely drinking water may mitigate low absolute lithium levels [24]. In addition, lithium appears to have value in augmenting antidepressant treatment. Lithium continues as the standard and most extensively evaluated treatment for bipolar disorder, especially for long-term prophylaxis [8, 14, 19].

Several epidemiologic studies have reported that lithium in drinking water may be associated with lower rates of suicide mortality [24–31], lower incidence of dementia, lower levels of adolescents' depression, aggression [32, 33] and psychotic experiences [34] at the population level.

The biological actions of lithium are complex and not fully understood. Lithium has been the subject of suicide research for a long time [7, 35, 36]. In addition to its mood-stabilizing, anti-depressive and anti-manic effects in individuals with bipolar and other major affective disorders, some studies suggest that lithium in therapeutic doses has an anti-suicidal effect [8, 14, 37– 39]. This may be mediated through its mood-stabilizing properties or through a reduction of aggressiveness and impulsivity, which are associated with increased risk of suicide [33].

Despite this interest, none of the studies, to the best of our knowledge has studied the interaction between the suicide rates, prevalence of mental and behavioral disorders and naturally occurring lithium in drinking water.

Based on the existing knowledge we hypothesized that higher levels of lithium in drinking water may exert an anti-suicidal effect in the population with high incidence of affective disorders.

AIM AND OBJECTIVES

The aim of this study was to establish whether suicide rates and

incidence of affective disorders are associated with lithium levels in drinking water.

Objectives of the study:

1. To examine the lithium levels in drinking water in Lithuanian municipal central wellfields.

2. To investigate the association of suicide mortality rates with lithium levels in drinking water.

3. To evaluate whether incidence of affective disorders is associated with lithium levels in drinking water.

Statements of the hypotheses

We hypothesized, that:

1. Lithium levels in drinking water differs across the municipalities; 2. Increased lithium levels in drinking water may be associated with a

lower suicide mortality rate;

3. Lithium levels in drinking water are negatively associated with suicide rates in the population with higher incidence of affective disorders.

Scientific novelty of the study

Firstly, this is the first ecological study on the association between lithium levels in the public drinking water and suicide rates in Eastern Europe. This study investigated the lithium level in drinking water wellfields across the Lithuanian municipalities, and tested the hypothesis that higher lithium levels in the public drinking water are associated with lower suicide mortality rates.

Secondly, we evaluated the association of mental and behavioral disorders with lithium level in drinking water. Our study revealed a higher incidence of affective disorder in municipalities with higher lithium level in drinking water. Therefore, it seems probable that the anti-suicidal effect of lithium may be unrelated to the mood-stabilising effects and that very low lithium levels (comparing with therapeutic doses) may possess an anti-suicidal effect. On the other hand, although lithium levels in drinking water are extremely low, long-term exposure to lithium may be a factor which mitigates low absolute levels.

The study results demonstrate a new knowledge in understanding the association of suicide rates with lithium levels in drinking water linked to

Finally, our study might provide new knowledge for future research studies evaluating biological mechanisms on suicidal behavior at the popula-tion level, especially in patients with affective disorders.

Contribution of the author

The author of the dissertation has been actively involved in the collection of water samples since the beginning of the study. She has been in direct contact with the heads of municipalities’ wells regarding the permission to collect water samples for lithium testing. In the course of the research, the author prepared a project to obtain funding for research of the central wells of Lithuanian municipalities. She also cooperated with the representatives of the Institute of Geology and Geography of Nature Research Center for the collection of water samples, and the representatives of the laboratory for the analysis of drinking water samples.

The author collected and processed socio-demographic, incidence and mortality indicators of Lithuanian municipalities, using the database of direct and indirect access to Lithuanian health indicators. She processed the results and performed statistical analysis and interpretation of the data.

During the doctoral studies, the author presented the results of the researches at international and national conferences in Lithuania and abroad. The obtained results were published in peer-reviewed foreign journals.

1. REVIEW OF LITERATURE

Globally, suicides are the second leading cause of premature mortality in individuals aged 15 to 29 years (preceded by traffic accidents), and number three in the age-group 15–44 years [40–42]. According to recent review by Bachmann et al (2018) crude and age-standardized suicide rates according to WHO regions, amount to 10.7 per 100,000 worldwide but varies in certain regions [43].

The Eastern Mediterranean region has suicide rates of 3.8 and 4.3, the African region 8.8 and 12.8, the Americas 9.6 and 9.1, the Western Pacific region 10.8 and 9.1, South East Asia 12.9 and 13.3, and Europe 14.1 and 11.9 (all crude and age-standardized) [44]. The highest suicide rates for both men and women are found in Europe, more particularly in Eastern Europe, in a group of countries that share similar historical and sociocultural characte-ristics, such as Estonia, Latvia, Lithuania and, to a lesser extent, Finland, Hungary and the Russian Federation [40]. Obviously, the European area presents the highest absolute or crude suicide rate, namely above the global suicide rate of 10.7 for both genders [44, 45]. This is the case despite the fact that since 1980, a drop in suicide rates was reached through preventive measures and assisted suicides were taken out of the statistics. On the other hand, however, data quality is much better in comparison to other regions of the world [46].

Although there has been some progress in reducing historically high mortality rates from suicide, it nevertheless remains an important cause of death, particularly among men [47]. It was estimated that nearly three times as many men as women die by suicide in high-income countries, in contrast to low- and middle-income countries, where the rate is more equal [48, 49].

It is very difficult to obtain reliable data on suicide rates. For example, assuring rates of suicide attempts is nearly impossible, not least because a suicide attempt may not come to anyone’s attention, much less to the attention of the health care system. Nevertheless, the registration of suicides and suicide attempts is a desirable goal towards better prevention, detection, and intervention [40].

1.1.2. Suicide in Lithuania

Although the suicide rate in Lithuania has been decreasing every year since 2000 [50], Lithuania is still of the top countries in terms of suicide rate. Lithuania ranks fourth in the world according to suicide rate, and it ranks second for the male suicide rate among all countries in the world [42, 44].

According to data provided by the Institute of Hygiene in Lithuania, 683 people committed suicide in 2018, down from 748 in 2017, 823 in 2016, 891 in 2015, 935 in 2014 and 1,085 in 2013 [50]. According to the Eurostat data, the average rate of death by suicide in the EU in 2016 was 12.85 (per 100,000 residents), while the SMR in Lithuania was 25.7 (per 100,000 residents) exceeding the European Union average two times [51, 52]. In 2018 it was at 24.4, which means it remained slowly decreased for two years [50].

The most vulnerable groups include mid-aged (45–59 years) and over 75 years’ old males living in rural area. Men are five times more likely than women to commit suicide, so municipalities are developing rapid intervention algorithms to address this trend [50, 52].

The Organization for Economic Cooperation and Development (OECD) notes that Lithuania’s high suicide rate is tied to many different factors, including rapid social and economic change which increase both psychological and social insecurity, as well as the lack of a national suicide prevention strategy [52].

1.1.3. Suicide prevention

Countries and communities may influence suicide rates by measures of primary and secondary prevention [1]. In recent years, the authorities have launched a number of suicide prevention campaigns. Social and economic conditions have also changed, leading to a 45% decrease in the number of deaths due to suicide between 2000 and 2016, but it is still more than double the OECD average for the general population and nearly three-times the OECD average for men [50, 52]. One of the goals of the Lithuanian Health Strategy 2014–2025 [53] is to decrease the SMR due to suicide down to 19.5 cases per 100,000 residents by 2020, and then to 12 cases by 2025. The rates

consumption. Measures included an increase in excise taxes, a prohibition of alcohol sales in some places such as petrol stations, an increase in the legal age for purchasing and consuming alcohol to 20 years, and a limitation of operating hours of sale [52].

Many strategies have been developed, not only to prevent suicide but also to detect depression symptoms earlier and provide more appropriate treatment for other mental health issues. Psychological help for those who are not willing or not able to seek out the professional assistance of a psychologist or a psychotherapist due to financial restrictions is available through online or phone suicide prevention hotlines.

In 2017, the National Audit Office renewed efforts to identify and support individuals at risk of mental health issues and to ensure immediate and continuous support to people who had attempted suicide, emphasizing the need for information sharing between institutions [52].

On 1st November 2018, the Lithuanian Health Minister published a paper stating the procedure for providing help to suicide survivors. The document includes prepared goals to increase training of key gatekeepers, volunteers, and professionals regarding recognition of risk factors, warning signs and at-risk behaviors; the provision of effective interventions; and the development and promotion of effective clinical and professional practice to support clients, families and communities. Psychosocial assessment can be a vital tool for self-harm management, engaging patients in treatment and improving their rates of aftercare.

A recent Lithuanian Health Minister order states that, from November 2019, new health care services will be provided in mental health care institu-tions to children and adolescents. In addition, the network of instituinstitu-tions pro-viding psychiatric day care treatment will expand from five institutions to ten.

Since 2019, the Ministry also started financing municipal Public Health Bureaus to promote mental health prevention in schools. The objectives of this program are to enhance the competences of school staff in detecting and addressing mental health issues and improving overall mental health literacy. According to data provided by the State Patients’ Fund, the number of visits to medical psychologists in clinics has tripled over the past three years – from 20,959 in 2014 to 65,000 in 2017. The support from other people during a suicidal crisis is one of the strongest suicide deterrents [55]. Unfortunately, individuals with suicide behavior often avoid seeking help from others [56].

There is also a problem with the reliability of suicide confirmation and reports that needs to be improved. Suicides are most commonly found misclassified according to the codes of the 10th _{edition of the International}

profound psychological and behavioral effects on individuals and families, who may avoid seeking help, and may conceal or deny suicide. This can cause false death registration and coding practice [58, 59]. Providing case registra-tion informaregistra-tion to policy makers, researchers and health professionals allows greater exposure of the problem of suicide and can be a way of raising awareness, initiating research and development in prevention campaigns, and monitoring the effectiveness of suicide prevention and intervention strategies. Until now, the reasons for the marked heterogeneity in international suicide rates are not fully understood. One important observation is that national suicide rates by particular lethal methods (such as hanging, poiso-ning, gassing, shooting, jumping, and drowning) vary greatly between nations [60], but tend to be stable within nations on a year-to-year basis [61]. This predictability of method specific suicide rates underpins most universal measures to prevent suicide.

Some potential universal prevention strategies come at a greater cost. Examples include better access to health care [62] and measures to reduce unemployment [63]. Other universal, potentially important measures might be inexpensive but hard to achieve, for example reducing suicide by reducing the stigma associated with accessing mental health care [64, 65].

Despite the challenges faced by Lithuania and other countries with a high suicide rate, universal measures hold the best hope for global suicide pre-vention.

1.1.4. Suicide and mental and behavioral disorders

Suicide is not a mental health problem itself, but is linked with mental distress. The etiology of suicidal behavior and suicidal ideation is multi-factorial, although one of the most common risk factors is having a psychiatric disease. Several psychological autopsy studies have supported high rates of psychiatric disorders among individuals who die by suicide [66]. Further, a meta-analysis of 3,275 suicides reported that 87.3% of suicide completers had been diagnosed with a psychiatric disorder prior to the suicide [67]. Other studies suggest that mental illness accounts for a large majority of suicides; numbers are at least 10 times as high as in the general population.

severe. One important feature of affective disorders is that they are associated with an extremely high risk of suicide, as the majority of those committing suicides are suffering from some type of affective disorder [72].

The suicide-related mortality among patients with affective disorders is approximately 30 times higher and overall mortality 2 to 3 times higher than suicide-related mortality in the general population [73]. In their meta-study of the mental health-suicide relationship, Ferrari et al. (2014) assess the pooled relative risk of suicide across a range of mental health and substance use disorders [70]. It has been reported that on average, 43.2% (SD 18.5%) of suicide cases were diagnosed with any affective disorders (including depressive and bipolar disorders) and 25.7% (SD 14.8%) with other substance problems [74].

Depression is the leading cause of death by suicide worldwide and the second cause of Disability Adjusted Life Years (DALYs) in the age category 15–44 years [70, 75]. Bradvik’s study (2018) showed that the majority of suicides are related to a psychiatric condition, including depression, substance use disorders, and psychosis [76]. Other research suggests that the suicide risk for mental disorders including depression, alcoholism, and schizophrenia is around 5% to 8%, while up to 60% of people who commit suicide have depression [77].

An individual with depression, for example, is 20 times more likely to die from suicide than someone without; someone with anxiety disorder around 3 times; schizophrenia around 13 times; bipolar disorder 6 times; and anorexia 8 times as likely. Although the total prevalence of mental health and substance use disorders does not show a direct relationship to suicide, there are notable links between specific types of mental health disorders and suicide.

Longitudinal studies have consistently demonstrated the importance of past suicide attempts, cigarette smoking, alcohol use disorders, and comorbid personality disorder as predictors of future suicide attempts in depressed populations [78, 79]. The important risk factor for lifetime suicide attempt is depression, which implies that the lifetime prevalence of suicide attempts could be reduced significantly by preventing depression and by recognizing and adequately treating it [1, 80].

Advances in understanding suicidal thinking, strongly intertwined with mood states, as well as psychological and neurocognitive factors influencing a person’s decision making [81, 82] are needed to advance the clinical care of suicidal patients and prevention of suicidal behavior.

Among the many risk factors for suicide are mental illness, physical illness, previous suicide attempt, substance abuse, family history of suicide, impulsiveness, hopelessness, isolation, loss, relationship, social, work, and financial [84–87]. Taking into account results from previous studies, it was hypothesized that the most people who encounter such challenges learn to cope with them or find ways to overcome them, going on to survive and sometimes flourish. However, some people with limited psychological reserves who face the same challenges might come to feel that suicide, however undesirable, is preferable to living [88, 89].

Although there is no simple explanation for such counter-intuitive human behavior, social and cultural factors, media exposure, and availability of lethal means are woven in a complex web with other risk factors that can lead to suicide. The complexity of risk factors for suicide suggests that many approaches to suicide prevention should be considered and customized to accommodate local circumstances.

1.2. The Anti-suicidal effect of lithium 1.2.1. Dietary source of lithium

Lithium is a trace element widely distributed on Earth [90, 91]. Because of its similarity to sodium and potassium, lithium easily crosses all biological barriers, meaning that it shows almost complete oral absorption and a uniform distribution in body fluids [91]. A study by Schrauzer reported that dietary

lithium which has received scant attention is found in grains and vegetables and to some extent animal derived foods [92]. In some areas, drinking water also provides significant amounts of the element. Water moving through the ground do react to varying degrees with the surrounding minerals (and other components), and it is these rock-water interactions that give the water its characteristic chemistry. Lithium is detectable at variable concentrations in drinking water throughout the world [8, 36, 93, 94]. Human dietary lithium intakes depend on location and on the type of foods consumed, and vary over a wide range.

In studies conducted between the 1970s to the 1990s, rats and goats maintained on low lithium rations were shown to exhibit higher mortalities as well as reproductive and behavioral abnormalities. In human defined lithium deficiency, diseases have not been characterized, but low lithium intakes from water supplies were associated with increased rates of suicides, homicides and the arrest rates for drug use and other crimes [95]. Although lithium does not have a known biological role and does not appear to be an essential element for humans, the available experimental evidence now appears to be sufficient to accept lithium as essential; a provisional RDA for a 70 kg adult of 1,000 µg/day is suggested [92].

1.2.2. The Effects of lithium deficiency on behavioral parameters

As lithium deficiency in humans is unlikely, any symptoms of deficiency, if at all observable, would be expected to be mild and manifest themselves primarily by behavioral rather than physiological abnormalities.

Evidence linking low lithium intakes with altered behavior and aggressiveness in humans was reported by Dawson et al. [96]. These authors compared the regional mental hospital admission and homicide rates for 1967–1969 with the lithium concentrations in tap water and urine samples obtained from 24 county sites in Texas. The highest significant inverse asso-ciations of water lithium levels were observed with first mental hospital admissions for psychosis, neurosis and personality disorders. The decreasing order of magnitude of the associations was neurosis, schizophrenia psychosis, first admission, all admissions, personality, homicide and secondary admis-sions. Lithium deficiency may not only be caused by low dietary lithium intakes but can also be secondary to certain diseases. Research shows that organ lithium contents of kidney disease and, especially, of dialysis patients are approaching deficiency levels [14].

1.2.3. Mechanistic considerations of lithium

Lithium acts on mood and suicidality via complex interactions with the serotoninergic system [67] to the decrease cerebral level of tryptophan and serotonin [66]. The research shows the important role testosterone plays in the regulation of mood and behavior, therefore it is a potentially important marker for suicide risk in an already at-risk population [67].

The biochemical mechanisms of action of lithium appear to be extra-ordinarily complex, multifactorial and strongly inter-correlated with the functions of other elements, drugs, enzymes, hormones, vitamins, growth and

unusually broad activity spectrum of lithium [11]. However, lithium has been shown to enhance folate and B12 transport into cells [67]. The transport of these factors is inhibited in lithium deficiency and can be restored by lithium supplementation. Since vitamin B12 and folate also affect mood associated parameters, the stimulation of the transport of these vitamins into brain cells by lithium may be cited as yet another mechanism of the anti-depressive, mood elevating and anti-aggressive actions of lithium at nutritional dosage levels. Recognition of the inter-correlated nature of all biological actions of lithium may result in improved therapeutic concepts. Thus, the joint admi-nistration of lithium with vitamin B12 and folate may prove more effective than lithium or the vitamins alone [14]. The fact that embryonal lithium concentrations are the highest during early fetal development suggests that it is specifically needed [96].

1.3. Summary of studies investigating the anti-suicidal effects of lithium in drinking water

Research has shown that lithium has preventive capabilities against suicide and even on a trace level, it can be therapeutic on other mental health issues. Multiple studies have attempted to find an inverse association between suicide and high trace-levels of lithium in drinking water.

The electronic search strategy on the PubMed database was conducted from the date of inception of the studies in the databases to October 2019. The following keyword combinations were used: “suicide” AND “lithium in drinking water”, or “lithium in public water”. Using the keyword combi-nations described above, the search in PubMed provided 42 results. Studies were included if they published in English, and investigating the relationship between the level of lithium in drinking water and suicides rates in the general population. After removal of 16 duplicates, of 10 reviews, of 2 opinions of the remaining 14 records were screened for eligibility.

1.3.1. Analysis of selected studies

[95]. Using data from 27 Texas counties for the period 1978–1987, it was found that the incidence rates of suicide, homicide, and rape were signifi-cantly higher in counties whose drinking water lithium levels ranged from 70 to 160 µg/L.

These results were replicated in two studies by Ohgami et al. [24] and by Kapusta et al. [26], one from Japan and one from Austria concluded that areas with higher lithium levels in the drinking water had lower suicide rates. They found that lithium levels were significantly and negatively associated with suicide standardized mortality ratio averages for the period 2002–2006 and suggested that even very low levels of lithium in drinking water may play a role in reducing suicide risk within the general population.

Knudsen et al. (2017) reported that lithium levels in the drinking water of 275 municipalities in Denmark ranged from 0.6 to 30.7 μg/L, and surpri-singly that there was an increasing suicide rate by increasing lithium levels [97]. These two studies had similar locations and lithium ranges but opposite findings.

However, it is notable that the significant relationship between lithium levels of drinking water and the suicide rate has been proven after adjustment for some confounders. Kapusta et al. [26] evaluated the association between local lithium levels in drinking water and suicide mortality at district level in Austria. The overall suicide rate as well as the suicide mortality ratio were inversely associated with lithium levels in drinking water and remained significant after sensitivity analyses and adjustment for socioeconomic factors.

However, Kabacs et al. [93], did not prove this association between lithium in drinking water and suicide rates across the East of England. The analyses of the data showed that there was no correlation between lithium levels in drinking water and suicide mortality rates in the 47 subdivisions of the East of England. This negative result could in part be explained by methodological weaknesses of the study, for example, sociodemographic or socioeconomic characteristics were not included.

y of st udi es in ve st igat ing the ant i-su ic id al ef fect s o f lith iu m in d ri nk ing w at er o f s amp les ar eas ) A dj us te d fo r Li th iu m le ve ls in dr inki ng w at er , µ g/ L R es ul ts s Po pu la tio n de ns ity 0– 16 0 Lo w li th iu m in ta ke s f ro m w at er su pp lie s w er e as so ci at ed w ith in cr eas ed rat es o f s ui ci des ip al iti es o f ct ure Po pu la tio n siz e 0. 7– 59 Ev en v er y lo w le ve ls of li th iu m in dr in ki ng w at er m ay p lay a ro le in re du ci ng su ic id e r isk w ith in th e ge ne ra l po pul at ion es fr om isions N o <1 –21 N o as so ci at io n be tw ee n lit hi um le ve ls in dr in ki ng (t ap) wa te r a nd m or ta lit y fro m su ic id e ts Po pu la tio n de ns ity , p er c ap ita in co m e, pr op or tio n of Ro m an Ca th ol ic s, m en ta l he al th se rv ic e pr ovi de rs 11. 3 (S D  ± 27 ) H ig he r n at ur al li th iu m conc ent ra tio ns in dr in ki ng w at er a re asso ci at ed w ith lo w er su ic id e m or ta lit y ra te s sam pl es pr ef ec tu re s Po pu la tio n de ns ity 11. 10 (S D ± 21 .1 6; 0. 1– 12 1) Ten den cy fo r l ow er su ic id e ra te s i n the p re fe ct ur es wi th hi gh le ve ls of lit hi um in d rin ki ng w at er iu m _ples, _ies Po pu lat io n den sit y, ag e, g en der , r ace/ et hni ci ty , m edia n inc om e pe r ho us eh ol d, pov er ty and u ne m pl oym ent ra te s 2. 8– 21 9 H ig he r l ith iu m le ve ls in th e pu bl ic dr in ki ng w at er w er e as so ci at ed w ith lo w er su ic id e ra te s ip al iti es , re fe ct ure , 373 ,3 39 ) Th e de ns ity o f m ed ic al in sti tu tio ns p er 10 ,0 00 pe opl e and the u ne m pl oym ent ra te 0– 12 .9 Si gn ifi can ce fo un d fo r f em al es be fo re a dj us tm en ts w ith a tr en d af te r adj us tm en ts (p = 0. 10) a nd not w ha tsoe ve r f or m al es

1. C ont inue d nces N umb er o f s amp les (n o. of ar eas ) A dj us te d fo r Li th iu m le ve ls in dr inki ng w at er , µ g/ L R es ul ts e t a l. us tri a 6, 46 0 lit hi um m eas ur es o f 99 d ist ric ts Po pu la tio n de ns ity , p er c ap ita in co m e, pr op or tio n of Ro m an Ca th ol ic s, de ns ity o f ps yc hia tris ts pe r 1 0, 000 p opula tio n, the num be r of g en er al p ract iti on er s, th e den si-ty of ps yc hot he ra pi sts pe r 1 0, 00 0 pe opl e 11. 3 (S D  ± 27 ) Su ic id e ra te , S M R in ve rs el y as so ci at ed w ith L i l ev el s e ve n af te r ad ju stm en t o f l ith iu m p re sc rip tio ns et a l. al y 14 5 sa m pl es Po pu la tio n siz e, m ount ai nou s a re a, hi ghl y ur ba ni ze d, ge ogr ap hi c loc at io n 5. 28 (0. 11 –6 0. 8) Li thi um c onc ent ra tio ns a nd loc al su ici de rat es w er e n ot si gn ifi can tly in ve rs el y re la te d l, 20 15 27 4 m un ic ip al iti es , K yu sh u Is la nd , Ja pa n, 2 01 1 Pr op or tio ns : o f e ld er ly p eop le , of 1 -pe rs on ho us eh ol ds , o f pe opl e wi th co lle ge e duc a-tion or m or e, o f pe opl e enga gi ng in pr im a-ry indus try , a nd une m pl oym ent ra te , a nnua l m ar riag e rat e, an nu al m ean tem pe ra tu re , an d an nu al p os tal sav in gs p er p er so n 0– 13 0 Si gn ifi ca nt ly in ve rs e as soc ia tions for th e ov er al l S M Rs (β = −0. 175, P = 0. 03 1) a nd for m al es (β = −0 .2 28 , P = 0. 005) , bu t no t f or fe m ale s ki e t a l. pa n 15 3 sa m pl es A nn ual to tal su ns hi ne, an nu al m ean te m pe ra tu re , a nn ua l t ot al ra in fa ll, a nn ua l to ta l s no w fa ll 3. 8 (S D ± 5 .3 ; 0. 1– 43 ) Li th iu m is in ve rs el y as so ci at ed w ith m al e bu t n ot fe m al e su ici de af ter ad ju stm en t o f m et eo ro lo gi ca l f ac to rs n et al, 27 5 m un ic ip al iti es in D en m ar k N o 0. 6– 30. 7 In cr eas in g su ic id e r at e by in cr eas in g lit hi um le ve ls e t a l, 54 P or tug ue se m un ic ip al iti es 2011 –20 16 Po pu la tio n de ns ity , i nc om e pe r c ap ita , une m pl oy m ent r at es a nd p rop or tio n of Ro m an Ca th ol ic s 10. 88 (S D 2 7. 18 ) N o as so ci at io n be tw ee n lit hi um in pu bl ic d rin ki ng w at er a nd su ic id e ra te s w as fou nd et a l, Fi ve dr inki ng wa te r sam pl es fr om each o f 15 A la ba m a count ie s Th e po ve rty h ad a p ar al le l t re nd w ith su ic id e ra te 0. 4– 32. 9 Si gni fic ant ly inve rs e as soc ia tions for th e ov er al l S M Rs a nd fo r m al es , b ut no t f or fem al es

Giotakos et al. [22] reported lower suicide rates in prefectures with high levels of lithium in drinking water. Analyses were conducted with respect to lithium levels in 34 prefectures of Greece from both rural and urban areas. The results indicated that there was a tendency for lower suicide rates in the prefectures with high levels of lithium in drinking water.

Confirmation of the association between levels of lithium in drinking water and suicide rate comes from a study published by Blüml et al. [27], which analysed 226 counties in Texas during 1999–2007. This study modeled the response of the county-level rate of suicide using both a linear and Poison rate regression adjusted for county-based population density, lithium levels, age, sex, race/ethnicity, median income per household, poverty and unemployment rates. Results showed that lithium levels were significantly associated with suicide rates.

A recent study in Italy by Pompili et al. [98] found that lithium levels in tap water and local suicide rates were not significantly inversely related. Based on lithium levels and analysed suicide data at the community level, and including the following covariates: totally mountainous areas, highly urbanized, and geographic location, were reported that suicide rates were not statistically significantly correlated with lithium levels.

The research group with Kapusta and colleagues [26] published a second study, Helbich et al. [25], on the same data set as that described previously, using a more refined statistical model based on the analysis of geospatial epidemiological data and lithium prescription. Previous studies have assumed that the lithium in drinking water originated from natural sources alone. They have not considered whether lithium prescribed to patients may have accumulated via waste water in groundwater aquifers after urinary excretion and interplay with natural lithium [23]. The results showed that lower lithium concentrations in the ground and drinking water might be responsible for higher suicides rates even after adjustment of lithium prescriptions.

The most recently published study on the issue, Shiotsuki et al. [30], evaluated the association between lithium levels in the public water supply and suicide rates after adjustment of meteorological factors. The results suggest that trace lithium is inversely associated with male but not female

increasing exposure to lithium through drinking water in Denmark. There-fore, we need randomized, placebo-controlled trials for determining the definite effect of lithium intake through drinking water on suicide incidence, although these studies may be the subject of ethical and clinical challenges.

Moreover, the cost-benefit analyses should be conducted to ensure that drinking water enrichment with lithium is both cost-effective and safe for vulnerable people such as children, pregnant women, and patients with special medical conditions.

Due to the different patterns of suicide attempts and suicide mortality among men and women, evaluating the relationship between drinking water lithium and suicide attempts, as well as other violent actions and impulsive behaviors, may be fruitful.

For example, in a systematic review and meta-analysis study Memon et al. [102] found that the protective (or inverse) association between lithium levels in drinking water and suicide mortality rates is likely to be stronger in populations with relatively higher suicide mortality rates.

1.3.2. Studies on potential anti-suicidal effects of lithium as a trace element in drinking water

Another approach to determine lithium’s suicide protective effects is to investigate the ecological association between suicide rates and lithium in drinking water.

Explication of the findings that even the very low levels of lithium provided in drinking water may reduce the risk of suicide is, at present, only speculative. It is possible, although unlikely, that the low levels of lithium such as those found in drinking water can bring about significant mood-stabilizing effects and reduce the risk of suicide in mood disorders not yet clinically defined by this mechanism [103]. In contrast, it could be speculated that even very low but long-lasting lithium exposure can enhance neurotro-phic mechanisms, neuroprotective factors and/or neurogenesis, which may account eventually for a reduced risk of suicide [26, 104].

It has been estimated that although lithium probably prevents about 250 suicide deaths per year in Germany, only 0.06% of the German population were prescribed lithium [39]. Similarly, a UK study showed that at least 40% of patients with bipolar disorder who died by suicide were not prescribed lithium or a mood stabilizer [105].

Several researches have shown that low-dose lithium augments the efficacy of antidepressant medications [106]. For those who are suffering

the risk of side effects and drug toxicity associated with high-dose lithium treatment. Low dose lithium has also demonstrated an ability to prevent and delay relapse in people suffering from alcoholism as well as stabilizing and improving mood in drug users [18].

Lithium also exerts powerful antioxidant and anti-inflammatory effects, thus helping to counter neurodegenerative diseases involving inflammation and oxidative damage [107]. The result is a healthier brain and reduced risk of neurodegeneration and mood instability.

While there is still a long way to go, and mental health still suffers more of a stigma than heart disease, for example, current research going on today, such as the mechanism by which lithium works, is an excellent step in the right direction [108]. It is an intriguing association, but not all articles are positive and the amount of exposure to lithium via drinking water is much, much lower than the exposure from lithium therapy.

Estimates of the population attributable fraction suggested that 12% (95% CI 4%–20%) of suicide-related events could have been avoided if patients had taken lithium during the entire follow-up [109]. Overall, the

findings summarized here, appear to provide strong and quite consistent support for the hypothesis that long term, but not acute, treatment with lithium may have a special role in reducing suicidal risks. It is important to acknowledge the significant role of mental disorders in suicide; underplaying it may result in missed opportunities for suicide prevention [110]. It seems we can step forward and conduct a series of clinical trials to test the hypothesis that long-term lithium intake through drinking water has a protective effect against suicide.

1.3.3. Summary of the literature review

To summarize the literature review: in terms of the studies observed, evidence suggests that there is an inverse relationship between lithium levels in drinking water and suicide mortality, overall. Significance in most the outcome indicators were found in 14 of 42 studies, partial evidence was established in 2 of them (Pompili et al and Sugawara et al) [28, 98] and none in three piece of research (Knudsen et al, Oliveira et al, Kabacs et al) [93, 97,

low in drinking water, long-term exposure to lithium may be a factor which mitigates low absolute levels.

Findings from a recent international multi-centre trial are compatible with the idea that long-term lithium treatment extends the survival of patients suffering from affective disorders to match the general population. Research from multiple countries, including the United States, has demonstrated a link between higher lithium levels in drinking water and lower rates of suicide in clinical and general populations [112].

It is still not clear what amount of lithium in drinking water can provide an independent protective effect for suicide. In particular, more research is needed to understand how lithium levels in the public drinking water supply correlate with blood lithium levels. Also, drinking water is not the only dietary source of lithium. According to the US Environment Protection Agency, some grains and vegetables are even richer in lithium than drinking water [92].

Overall, the findings summarized here, appear to provide strong and quite consistent support for the hypothesis that long term, but not acute treatment with lithium may have a special role in reducing suicidal risks. It is important to acknowledge the significant role of mental disorders in suicide; under-playing it may result in missed opportunities for suicide prevention [110].

Most evidence suggests that the negative association is present, but the heterogeneity of the studies and the lack of essential considerations in many of them make it hard to conclude such a connection is present. More needs to be done to shed light on an understanding of lithium and its effect in a low dose, as well.

2. MATERIAL AND METHODS

The research was designed as an ecological study using cross-sectional data gathered in municipalities (cities or district’s centres) of Lithuania. This study was performed at an aggregate level (no individual-level) to suggest an association between suicide mortality rate and lithium levels in drinking water. The study consists of two stages (Fig. 2.1.1).

The unit of analysis was the municipality (district’s centres or cities). This ecological study was designed by aggregating suicide mortality rate (outcome) for each studied municipality for a five-year period and were analysed using linear regression with lithium levels in drinking water in the central wellfields of municipalities as exposure.

2.2. Methods

2.2.1. Study area: evaluation of lithium levels in drinking water

For the analyses of lithium concentration in drinking water, we collected groundwater samples in the central wellfields of Lithuanian municipalities, which water supply systems serves the largest percentage of drinking water customers.

Study I (Pilot study)

Firstly, in the pilot study, we focused on the urban areas but not on the rural areas in order to decrease as much as possible the impact of the heterogeneity of social, economic and cultural background. In 2013 Lithuania had 2,971,905 inhabitants, 1,367,076 men and 1,604,828 women. There are 103 cities, 249 towns, and 21,000 villages in Lithuania [113] .

The country’s 5 largest cities – Vilnius, Klaipėda, Kaunas, Šiauliai, Alytus (one exception was Panevėžys, because no response has been received from the department of wellfield) and 4 resort cities – Palanga, Neringa, Birštonas, Druskininkai had the population of 1,190,261 (41% of the total population of Lithuania, with a range from 2,525 in Birštonas to 526,356 in the capital city Vilnius [113].

This pilot study was conducted over a period of 3-month (November 2013–January 2014). Twenty-two samples (range from 1 to 5 per city) of drinking water from all (22) wellfields were taken in cities municipalities (5 samples – Vilnius, 4 samples – Kaunas, 3 samples – Klaipėda, 2 samples – Šiauliai, 2 samples – Alytus, 2 samples – Birštonas; 2 samples – Neringa, 1 sample – Druskininkai, and 1 sample – Palanga).

The drinking water samples were analysed for lithium concentration by inductively coupled plasma mass spectrometry (ICP-MS) NexION 300D (Perkin Elmer, USA). The lithium detection limit, using seven replicated spiked with 1.0 µg/L of Multi-Element Calibration Standard 3 (Pure Plus PerkinElmer, USA) was 0.154 ± 0.03 µg/L and derived as described [114].

Study II

Klaipėda, Šiauliai, Panevėžys and Alytus) are two different municipalities: municipality on the city and district.

In the second study (Study II), over a period of 2-month (June 2017-July 2017), we collected 56 drinking water samples (1 sample per municipality, with one exception – 3 samples in Klaipėda municipality) from 56 central wellfields of the country’s municipalities, including cities municipalities (Druskininkai, Birštonas, Klaipėda, Neringa, Palanga, Kalvarija, Kazlų Rūda, Marijampolė, Rietavas, Visaginas, Elektrėnai) and districts municipalities (Alytus, Lazdijai, Varėna, Jonava, Kaišiadorys, Kaunas, Klaipėda, Kėdainiai, Prienai, Raseiniai, Klaipėda, Kretinga, Skuodas, Šilutė, Šakiai, Vilkaviškis, Biržai, Kupiškis, Panevėžys, Pasvalys, Rokiškis, Akmenė, Joniškis, Kelmė, Pakruojis, Radviliškis, Šiauliai, Jurbarkas, Šilalė, Tauragė, Mažeikiai, Plun-gė, Telšiai, Anykščiai, Ignalina, Molėtai, Utena, Zarasai, Šalčininkai, Širvin-tai, Švenčionys, Trakai, Ukmergė, Vilnius).

In Klaipėda, city and district municipalities were taken 2 more drinking water samples (overall 3 samples in 3 wellfields) for assessment of different aquifers, which are exploited in drinking water supply systems. The mean value was calculated for analysis.

The lithium concentration in these samples was determined by the ion chromatography DIONEX ICS-1000 (Thermo Scientific, USA) employing the standard LST EN ISO 14911.

We cooperated with geologists and geoscientists from the Institute of Geology and Geography of Nature Research Centre in collecting drinking water samples and analysing the lithium concentration.

2.2.2. Dependent variable: suicide standardized mortality rate

We obtained standardized mortality rate (SMR) for suicide (per 100,000 population) from the Health Information Centre of the Institute of Hygiene (Lithuania Database of Health Indicators) and used the average suicide SMR for 2009–2013 (for each of 9 municipalities) and 2012–2016 (for each of 53 municipalities). In accordance with International Statistical Classification of Diseases (ICD-10) the data included causes of death with codes X60-X84 [50, 115]. SMR were age standardized using the European Standard Population

per 1,000, women/men proportion (number of women per 1,000 men) and were obtained from the Department of Statistics, and from the Institute of Hygiene database, and were averaged across the investigated time period for each city or district’s municipality, respectively.

In an addition to the municipal socio-demographic characteristics, morbidity indicators such as incidence (per 100,000) of mental and behavioral disorders (MBD) (ICD-10 codes F00-F99), incidence of diseases of the ner-vous systems (ICD-10 codes G00-99), number of suicide attempts (ICD-10 codes X60-84), and antidepressants usage rate across all age groups and gender within the five-year period (2012-2016) were included in the final analysis as an adjustment factors. Incidence of MBD included affective disorders (ICD-10 codes F30-39), schizophrenia (ICD-10 codes F20-29), MBD due to use of alcohol (ICD-10 codes F10).

The analysis of antidepressant use was based on the Defined Daily Dose (DDD) that is most informative in general terms as respondent based methods create difficulty with estimating the level of use. The DDD based approach does not capture most aspects of clinical practice but provides an objective record of overall level of medicine use. Because wholesalers do not accumulate drug reserves, the DDD methodology provides the most adequate representation of drug use at the level of actual sales in the whole country. While the purchased drugs are not consumed entirely, there are no grounds to believe that failure of compliance at this level was different between demographically similar regions.

2.3. Statistical analysis

All statistical analyses were performed using software from the Statistical Package for Social Sciences 17.0 for Windows (SPSS, Inc., Chicago, Ill., USA). A value of P<0.05 was considered significant.

A descriptive analysis was conducted to describe the profile of the municipalities. Tests for normality were conducted using a Kolmogorov-Smirnov test and a Pearson correlation coefficients or Spearman rank correlation coefficients were calculated between original and transformed variables of lithium and standardized suicide mortality rates, and among the municipalities’ routinely collected data. A curve estimation regression analy-sis was used in SPSS examining an exponential or quadratic relationship.

Because of significant differences in population size across the munici-palities, weighted least squares regression analysis, adjusted for the size of each population, was used to investigate the association between suicide

Study I (Pilot study)

At first, we transformed a nonlinear model to a linear model, which can be analysed using Linear Regression procedure. When a nonlinear pattern was corrected by a log transformation, it was accompanied by a substantial increase in the value of R²: from 0.068 (P = 0.248) to 0.467 (P = 0.025). If the value of R² changed, then the relationship was not linear and the transformation was effective. The addition of the logarithmic term produced impressive results indicating that there is a strong relationship between suicide SMR and lithium levels, when lithium level is represented by a combination of the original variable and the log form of the original variable (R2 _{= 0.774, P = 0.005).}

Multivariate regression models incorporated the number of women for 1,000 men as covariate.

Study II

The least squares linear regression modelling weighted for local population size and including nonlinear terms into linear regression were performed. The most common non-linear term is the quadratic term constructing a regression model with the predictor and predictor2 _{terms. The} quadratic term (X2_{) was included in the linear regression model because of} the inverted U-shape relation of lithium to suicide SMR. To avoid a multicollinearity problem with the original variable [117] and its quadratic term (X2_{), a predictor variable (lithium) was centered (subtracting the mean)} to create the square term:

Xc = (Lithium – mean) and Xc2 =(Lithium – mean)2.

The significance of quadratic terms could signal that the relationship is non-linear. The sign merely represents the type of non-linearity. A positive quadratic term could suggest that the relation is exponential. A negative relation suggests that for high values the relation becomes negative.

The curve does not need to contain both sides of the U. A quadratic (squared) term turns a linear regression model into a curve. But because it is X that is squared, not the Beta coefficient, it still qualifies as a linear model (“linear regression” means linear in the Beta coefficients).

We initiated a crude model analysis of the association of lithium levels in drinking water and suicide SMRs without any adjustment of the confounding factors. Please note the sign for x2_{in each of the models. The}

sign is negative when the curve is concave (apex at the top, curve opens down).

The associations of lithium levels in drinking water with suicide SMRs (total, men and women) were investigated adjusting for unemployment rate, visits to psychiatrists, divorce rate, and women/men proportion. Adjusted models include covariates with P<0.1 in a univariate analysis, apart from proportion of women per 1,000 men, which was correlated to population size.

Analysis according to low and high lithium exposure groups

Common statistical techniques multiple linear regression and ANCOVA were used and R-squared was used as the effect size.

The mean lithium levels in drinking water were categorized into quartiles: Q1<3.5 μg/L, Q2 = 3.51–7.0 μg/L, Q3 = 7.1–20.0 μg/L, and Q4>20 to 39 μg/L. The municipalities were assigned to study groups on the basis of a median split. The median lithium concentration was calculated (7.0 μg/L) and municipalities with lithium concentration in drinking water greater than the median (N = 26) were assigned to the high lithium exposure group, the remaining municipalities (N = 27) were assigned to the low lithium exposure group. All analyses were repeated separately in these groups.

We estimated the minimum sample sizes required for multiple linear regressions and ANCOVA by using Power and Sample Size software based on the pre-specified values of alpha, power and effect size (R2_{). A larger} sample size is required for a smaller value of R-squared and when there are many explanatory (or independent) variables are to be tested in the regression model [118]. Sample size for a value of R-squared which is more than 0.4 the minimum sample size based on number of tested variables with selected R2 (Alpha = 0.05 and Power = 0.8) required will usually be less than 26 for a maximum of 5 independent variables [118].

Crude and multiple linear regression analysis were performed using the suicide SMRs (total, men, women) as the dependent variable. Multiple regres-sion analyses were performed using forward or backward procedures, which first treated lithium levels as an independent variable, and then included municipalities’ characteristics as the independent variables.

Because a bivariate correlation analysis of the independent variables showed potential for multicollinearity, Variance Inflation Factors (VIFs) were investigated. If all VIF scores were below the critical value of 5, as suggested by Fox (2002), multicollinearity was rejected [119]. The F-test with

As the number of confounders was large, some may feature multicolli-nearity. Stepwise multiple regression analysis was used to build a regression model with minimum collinearity between the variables. Since the impact of individual confounders was of secondary interest, principal component analysis (PCA) was employed to obtain a reduced set of principal components (PC) that were uncorrelated and depicted the overall risk factor variability. Factor scores were generated using the Regression Method. The scatter-plots for visualization of relationships between variables were used. Multiple linear regressions were repeated with the reduced set of new variables (principal components) as the independent variables.

3. RESULTS

3.1. Lithium levels in drinking water, suicide SMRs, and socio-demographic characteristics in the municipalities

Study I (Pilot study)

In total, 22 drinking water samples from the wellfields of 9 municipalities were analysed for lithium levels. Table 3.1.1 lists the municipalities’ socio-demographics characteristics, suicide SMR, and lithium level. Overall, for 5-year period (2009–2013), the average of suicide SMR in nine municipalities was 27 (range 16–50) per 100,000 of the population, 51 (range 29–93) per 100,000 of the population for men, and 7 (range 0–13) per 100,000 of the population for women.

The average was 2.4 samples per municipality, with a range from 1 to 5 samples. The mean lithium level, as evidenced by raw values for 22 samples, was 10.9 (SD 9.1) μg/L ranged from 0.48 to 35.53 µg/L while median level was 3.6 µg/L.

Table 3.1.1. Suicide SMRs, lithium levels, and socio-demographic indicators

in the municipalities (N = 9)

Municipalities Population

in 2013 Women/ mena Suicide SMR

b _Lithium,

µg/Lc

Total Women Men

Neringa 2,719 1,013 50.36 13.3 93.6 1.24 Klaipėda 158,541 1,205 19.33 5.7 36.3 13.1 Palanga 15,732 1,257 19.05 4.9 37.3 21.79 Šiauliai 106,470 1,304 26.88 9.2 50.1 28.68 Kaunas 306,888 1,269 21.03 9.3 36.1 11.63 Druskininkai 14,128 1,267 29.09 7.6 55.6 3.89 Birštonas 2,525 1,243 29.56 0 64.9 5.05 Vilnius 526,356 1,232 15.97 5.9 29.2 5.87 Alytus 57,281 1,177 29.7 7.2 56.8 6.94

a _{number of women for 1,000 men;} b _{SMR, standardized mortality rate per 100,000 (European}

standard population); c _{average of lithium level.}

There were municipal differences in suicide SMR of approximately 3.2 times for men (from 29.2 to 93.6 per 100,000) and approximately 13 times for women (from 0 to 13.3 per 100,000).

Study II

The lithium levels in drinking water across 53 municipalities are shown in Fig. 3.1.2. The mean of lithium concentration across municipalities was 11.5 (SD 9.9) µg/L ranging from 1.0 to 39.0 µg/L, median – 7.0 (IQR 3.5– 20) µg/L.

Fig. 3.1.2. Lithium levels in drinking water samples collected

from central wellfields of municipalities

Six regions with different lithium levels in drinking water were distin-guished (Fig. 3.1.3). The Eastern Lithuania, Žemaitija and Šilutė had minimal values. While the regions of Central Lithuania, Northern Lithuania and Klaipėda had maximal values. These levels are obviously related to the aquifers of the corresponding lithological composition.

Fig. 3.1.3. Distribution of lithium levels in drinking water across Lithuania

Descriptive statistics of suicide SMRs and municipal socio-demographic characteristics registered within the five-year period (2012–2016) are presen-ted in Table 3.1.2.

Table 3.1.2. Suicide SMRs, lithium levels, and socio-demographic indicators

in the municipalities (N = 53)

Minimum Maximum Mean Std. Deviation

Suicide SMR total 16.25 65.50 39.54 10.96

Suicide SMR men 31.89 137.09 79.05 22.17

Suicide SMR women 3.76 28.79 13.28 5.13

Unemployment rate, % 5.50 17.14 11.32 2.93

Divorce rate per 1,000 2.35 4.74 3.19 0.46

Visits to psychiatrist, n/100 7.27 55.14 27.46 8.17

Women/mena _{972.40 1,275.40 1,135.87 47.85}

SMR, standardized mortality rate per 100,000 (European standard population); a _{number of}