Demographic risk factors increasing risk of seeking psychiatric emergency care after heat exposure

(1)

SAHLGRENSKA ACADEMY

Demographic risk factors increasing risk of seeking psychiatric

emergency care after heat exposure

Degree Project in Medicine

Joel Jageklint

Läkarprogrammet

Gothenburg, Sweden 2020

Supervisor: Steinn Steingrimsson, MD, PhD

,

(2)

2

Abstract

Project title: Demographic risk factors increasing risk of seeking psychiatric emergency care after heat exposure

Degree project, Programme in medicine, by Joel Jageklint. Supervisor: Steinn Steingrimsson and Hanne Krage Carlsen.

Background: In the last decade it has become apparent that ambient heat effects mental health negatively, but the demographics of which mental health patients are at risk or the specific risk factors predisposing mental health exacerbation during high ambient temperatures are less understood.

Aim: To identify specific groups of patients that have increased risk of seeking psychiatric emergency care during high ambient temperatures, and low ambient temperature as a secondary outcome.

Method: A time stratified case-crossover method was used comparing the number of visits to a single psychiatric emergency ward during hot or cold days (daily average temperatures above the 95th or under the 5th percentile) to days where temperature was calculated to have a minimum effect on the number of visits. A timeseries-model was used which account for daily, monthly and yearly variability and takes into account delayed effects in time from temperature. Analysis were performed separately for total patients, males and females, age groups 18-34 and 35-64, psychiatric outpatients and non-outpatients, and for patients with substance use disorders.

(4)

4

Conclusion: The findings of this study suggest that males and young adults (aged 18-34) are at highest risk of seeking psychiatric emergency care during high ambient temperatures. Psychiatric outpatients do not have an increased risk of seeking psychiatric emergency care during heat, while patients with substance use disorder might have an increased risk. In addition, cold temperatures seem to have an adverse effect on mental health in males and in young adults.

(5)

5

Introduction

Heat and adverse health outcomes

In 1995 an intense heatwave struck Chicago which claimed more than 600 lives. (1) In 2003 another heat wave struck Europe, where excess mortality for all of Europe has been estimated up to 70 000 (2). While it was previously known that high temperatures could have an adverse effect on health, the magnitude of the impact of heat was likely underestimated. Since the heatwaves of 1995 and 2003, combined with the increased awareness of the reality of global warming and climate change, the dangers of high ambient temperatures have become more understood. Since the beginning of the 21st century, the amount of research conducted on the topic of global warming or adverse weather events such as heat waves and their effect on public health has largely increased. As temperatures across the globe continues to rise and heat waves become more frequent (3)(4),the need for further understanding of the health effects caused by high temperatures become more and more important, because of the need for individuals, society and healthcare to be able to combat and adapt to such changes.

Research has found that individuals with renal disease, ischemic heart disease, and mental illness are among those who are negatively affected during heatwaves. (5). While the reason for why patients with renal insufficiency or heart disease show an increased risk for adverse health events have might be intuitive, as they are more susceptible to dehydration or hypotension, the reason for why patients with mental illnesses are at an increased risk would seem much less intuitive. While other groups had been previously been identified as risk groups during heat, the amount of research on the vulnerability of the mentally ill during high temperatures was very scarce entering the 21st century.

Suicide and heat

(6)

6

out during the spring and summer season, and peaking during the summer. (6) There also have been studies that showed that specifically the number of violent suicides increase during the warmer seasons. (7) This coincides with recent research that shows an increased agitation and number of assaults during the summer (8). It doesn’t seem far-fetched then to think that heat could exacerbate other mental illnesses, as those at risk of suicide often suffer from an underlying psychiatric condition, the most prominent being depression.

In recent studies, suicide has been shown to increase as much as 33% at 27 °C compared to average daily temperatures of the first percentile. These numbers are from a study by Kim et al, (9) who studied 1 320 148 suicides. Another study have shown an increased risk of death by suicide in the U.S. by 0.7% per 1 °C and a 2.1% increase in Mexico per 1 °C increase in average temperatures compared to monthly average temperatures. (10), and yet another in England and Wales observed an increased risk of suicide of 3.8% per 1°C above 18 °C in England and Wales (7).

Mental health impacted by high temperatures

While the relationship between suicide and heat was previously studied to some extent, other mental illnesses in relation to heat was not widely studied before the 21st century and was therefore a relatively unknown subject. However, much research has since been conducted on the subject.

(7)

7

mental and behavioural disorders not including senility, organic (including symptomatic) mental disorders, dementia, mood affective disorders, neurotic, stress related and somatoform disorders, disorders of psychological development, and senility. When analysing mortality specifically they also saw an increased incidence risk ratio for some groups of patients. This was seen among the following: those with any mental or behaviour disorder aged 65-75, patients with dementia aged 15-64 and even higher among males in that age group, disorders due to psychoactive substance use among females all ages, even higher among females 15-64 years. And lastly for patients with schizophrenia, schizotypal, or delusional disorder of all ages, the risk peaking among males and for those at or above 75 years of age.

This study brought forth significant evidence that patients suffering from psychiatric conditions were more likely to be admitted to hospitals during heatwaves and that some groups of patients with psychiatric conditions even had a higher mortality rate during heat waves. Following this research by Hansen et. al, many articles have since emerged on the topic of mental health and heat, as evidence pointed towards a broader group of psychiatric patients being adversely affected by heat, not only those patients at risk of suicide.

Another early study from Page et al. (2012) (12) studied mortalities in relation to temperature with data from a national register of deceased primary care patients, which include primary diagnoses of patients. This study was conducted in England, having a much milder climate compared to Australia. Results showed an increased risk of death by 4.9% per 1°C above 18 °C, (the 93rd percentile of annual temperature). On studying risks of specific patient groups, they found that those under 65 years of age had a higher risk of death at 10%, and those diagnosed with substance misuse not including alcohol had a risk of 20%. They also found that patients prescribed anti-psychotics or hypnotics/anxiolytics also had a higher risk of death, at 7% and 8% respectively.

(8)

8

health. Data gathered for this study regarded mental health issues, such as stress, depression or other mental disorders. Data from 2 million individuals were included in this study, from the time period of 2002-2012. The data was then analysed in relation to temperature, multi-year warming, and for natural disasters. They found that individuals over had an increased risk of scoring 0.5% higher for mental problems when monthly temperature averages shifted from 25-30 °C to higher than 25-30 °C. When stratified by sex no significant association was seen for men, while women showed an increased risk of 41%. They also found that per 1°C increase in average temperatures over a five-year period was associated with an increased prevalence of mental health issues by 2%.

This study did account for specific diagnoses, but rather a broader scope of mental-health related issues. The study neither takes into account the severity of the mental health issue. As the authors states however, this data could include cases of both clinical and subclinical manifestations of mental health issues, which are rarely accounted for when using more traditional study methods. This could include many individuals who are adversely affected by heat but do not seek medical help. Even if many individuals included in this study potentially did not have a medical diagnosis, it should still somewhat accurately reflect the response of the population’s emotional wellbeing to increased temperatures.

These are some of the studies that show that heat could have a powerful negative impact on those suffering from psychiatric disorders, even possibly leading to the death of a patient. The point can then be made that it is of critical importance to further research specifically which patients are at risk and find methods to protect or even properly treat these patients during days or periods of high ambient heat.

Temperature effects are often delayed in time

(9)

9

exposure. Lag 1 would mean the day after the day of exposure, while lag 0 would mean the actual day of exposure. A lag period is a specified amount of days following exposure that the desired outcome is measured. For example, lag 0-3 would mean the period including of the day of the exposure up until the third day following the day of exposure. By using this method, the outcome of each day following exposure is analysed. The researcher could then observe the effect during specific days (i.e. lags) or calculate the cumulative effect of several days (i.e. a lag period). Several studies conducted on mental health and temperature have since used this lagged models to better estimate the total impact temperature could have on mental health (17) (18).

Adaptation to temperature

The temperature threshold at which risk increases seems to vary greatly. In the study from Page et al. in England (12) they observed a threshold at 18 °C at which mortality rates increased among some psychiatric disorders. In Australia however, Hansen et al. observed a threshold for hospital admissions at 26.7 °C (11). A study conducted in China by Liu et al. (19) reported increased risk for emergency visits for mental health problems at 35 °C, and a study by Carlsen et al. in Sweden (18) reported increased risks of psychiatric emergency visits at 21.4°C. By this observation it is clear that the temperature at which increased risk for mental health related issues is reported vary greatly between locations where studies are performed.

While the focus of the following study was not on mental health, but rather on overall mortality in relation to temperature, Nordio et al. (20) compared mortality risks between different cities. Their method was manually grouping 211 cities in the U.S. into different clusters, to achieve a homogeneity within the cluster regarding temperature and air humidity. They then calculated risk ratios of mortality between days with 15.6 °C and 26.7 °C within the clusters. When comparing the risk ratio between clusters they found that clusters (i.e. cities) with a higher average temperature had a lower risk ratio of mortality at 26.7 °C than cities with a lower average temperature.

(10)

10

sudden increase in temperature compared to the local average temperature, no matter the baseline, that affects a population’s health negatively. This is strengthened by the numerous scientific articles that uses different temperature thresholds at which adverse effects are measured, but still observe a significant impact. It is plausible then that these differences depend on the lack of acclimatisation or adaptation. Adaptations to higher heat could be dependent on the use of air conditioning, or people learning to stay indoors during the peak of the day, or other habitual adaptations or even physiological acclimatisation. To strengthen this argument, Nordio et al. mentions several studies that show that despite globally increasing temperatures, overall heat mortality has actually decreased (21,22).

One of those studies mentioned was conducted in Stockholm by Oudin et al. (22), studying overall mortality rates during days with temperatures over the 98th_{and under the 2}nd_percentile

for the last century, 1901-2009. A linearly declining trend in mortality could be seen for the whole study period. This also suggests some form of adaptation, even though the reason for such adaptation is difficult to determine. However, it should also be noted that public health improvements could not be accounted for in this study, which could also contribute to lower mortality rates.

From these studies we cannot determine if such a decline in mortality have occurred for mental health-related illnesses. It might be of importance however to consider that the effect heat might have on mental health could be caused by sudden changes in temperature that individuals have not yet adapted to, either habitually or physiologically, and that a specific temperature value in one country might produce very different results in another country with a different climate. In most research articles concerning mental health and temperature, specific percentiles of the average annual or summer temperature are often used, such as temperatures at the 90th or 95th percentile rather than set temperature values.

Emergency visits related to mental health during high ambient temperatures

(11)

11

instead of mortality, one is more likely to include mental health-related issues that are not life-threatening, but which might very well impact the life of the patient negatively. By this method one will also include patients who have not previously had any mental illness but that might experience an onset of mental illness related to extreme temperatures. This contrasts with analysing mental-health related mortality, where a patient would need a psychiatric diagnosis before the event to be included in the study population.

The following studies are some but not all of studies who have studied the number of psychiatric emergency visits in relation to heat. Such studies have been conducted in Toronto, Canada (17), Quebec, Canada, (23), Gothenburg, Sweden (18), California U.S. (8), Jinan, China (19), and South Korea (24), all showing evidence of increased rates of emergency room admissions during periods of high ambient temperature.

In Toronto, Wang et al. studied emergency room admissions related to mental and behavioural disorders (MBDs), and the cumulative lagged effect of high ambient temperature on mental health (17), being among the first to study the lagged effect. Wang et al. found that for four days following a heat event, written as lags 0-4, a statistically significant effect could be seen on the risk of patients with MBDs being admitted to an emergency room. A heat event was defined as mean daily temperature at the 99th percentile (at 28 °C), with temperature at 50th percentile (8.7°C ) as reference. At the first day of a heat event, a risk ratio of 1.06 was observed for MBDs. Over the lag period of 0-7 days, a cumulative risk ratio of 1.29 was observed for MBDs.

Vida et al. showed that in Quebec, Canada, (25) a significant increase in the amount of mental health emergency department visits could be seen in two out of three areas studied when temperatures reached an average of 22.5 and/or 25 °C.

(12)

12

and cold seasons, and significant results were seen both for warm and cold seasons, meaning sudden increases in temperature even during cold seasons were seen to influence the risk.

In Jinan, China, Liu et al (19) observed an increased odds ratio of 2.231, 2.836, 3.178 and 2.988 for the odds of hospital visits related to mental health. The study was conducted during four heatwaves during the summer of 2010. The odds ratios were observed for 0-2 or 0-3 days following a heat wave. The heat waves were defined as three consecutive days with daily maximum temperature of 35 °C or higher.

In Gothenburg, Sweden, a study by Carlsen et al. (18) was conducted on the number of visits to a psychiatric emergency ward using a lag model. The study period spanned 2012-2017, and they analysed both the summer seasons of May to August and the winter seasons of November to February. The lag periods used in this study was 0-3 days and 0-14 days. At temperatures of the 95th_{percentile (21.4°C) during the summer season, they saw a cumulative increased risk of}

14% for lags 0-3, and 22% for lags 0-14. During the cold season they did not find statistically significant results. They also performed analysis on the 5th percentile, where a trend towards higher risk could be seen during the cold season, although it was not statistically significant.

Demographic or diagnostic risk factors

As the association between heat and psychiatric emergency visits almost seems to be unanimous, another important aspect moving forward is to research which specific groups of patients are affected by heat, since mental illnesses cover a very broad range of illnesses and demographics. By identifying which groups are at higher risk, psychiatric emergency wards could be better prepared in the future to care or treat these groups of patients. It will also help future research focus on specific groups of patients when researching underlying mechanisms of exacerbation by heat in patients with psychiatric conditions in order to further improve health among these individuals. It might also be possible to work proactively with certain groups at risk to prevent an exacerbation or possibly even onset of a psychiatric condition.

(13)

13

mood disorders for lags 1-3, with a cumulative risk ratio of 1.68 over 0-7 days, and for schizophrenia and schizotypal and delusional disorders during day 0-4, with a cumulative risk ratio of 2.49 over 0-7 days. In Quebec, Vida et al (25) found that both patients above and under 65 years of age had an increased risk but did not further study specific diagnoses. The study by Liu et al. (19) in Jinan, China did not analyse specific diagnoses, but did point out that elderly, outdoor workers, singles, and people living in urban areas showed a statistically significant higher odds ratio. Elderly: 3.034, outdoor worker: 1.714, single: 1.709, and living in an urban area: 1.523.

Basu et al. (8) found that when stratified by age, the highest risk increase was observed in 6-18-year olds, having a 7.3% increased risk. This study also studied self-inflicted injuries or suicides which also had the highest risk among 6-18-year olds. Patients under 18 are not commonly included in studies on this subject, and as such it is interesting to note that this patient group might also be affected by heat.

In the study by Page et al. (12) found an increased risk of mortality during high temperatures for those with other substance use not including alcohol by 20%, and for those under 65 years of age at 10%, as well patient prescribed anti-psychotics or hypnotics/anxiolytics at 7% and 8% respectively. This indicates that prescribed medication could also be involved in mediating the effect between heat and health of those with conditions of mental illness. However, compliance to medications could not be confirmed in this study.

Overall, there appears to be several diagnostic or demographic factors which could predispose for a patient being vulnerable at high temperatures. Which groups of patients are vulnerable often vary between studies, and there is so far a lack of consensus because of varying evidence as to which groups of patients are at risk.

Introduction to the current study

(14)

14

those patients who already have an underlying psychiatric disorder are at an increased risk of visiting a psychiatric emergency ward, or whether the risk is increased for those who do not have such an underlying disorder, arguably even indicating a new onset of a psychiatric disorder. Furthermore, this study will also study patients with substance use disorders, as some evidence have previously shown an increased risk among these patients. (12) An explanation so simple as more outdoor restaurants opening during summertime in Sweden, and people spending more time outdoors and being free from work could hypothetically lead to an increased consumption of alcohol or other substances, and one of the aims of the following study will be to determine the risk of this patient group.

The relationship between cold temperatures and mental health have not been extensively studied, as most studies focus on the effect from high temperatures. The current study will also aim to determine whether cold temperatures during the summer season could exert an effect on mental health as a secondary outcome.

Aim

The aim of the study is to assess whether certain demographic or diagnostic patient groups have a higher risk of visiting a psychiatric emergency ward located in Gothenburg, Sweden, during hot periods as a primary outcome and cold periods as a secondary outcome. The groups of patients studied will be total number of patients, males, females, patients aged 18-34, 35-64 and 65 years or older, patients with substance use disorder, patients who are a current psychiatric outpatient and patients who are not a current psychiatric outpatient. Each group will be studied independently of each other to estimate risk ratios for each group of patients.

(15)

15

Materials and Methods

Data for temperature were gathered from Swedish Metrological and Hydrological Institute (SMHI), from an online open databank containing daily mean temperatures (available online:

https://www.smhi.se/data/meteorologi/ladda-ner-meteorologiska-observationer#param=airtemperatureInstant,stations=all, extracted 2020-03-03). The specific

station from which temperature was gathered was Göteborg A, a measuring station in the centre of the city located 3 metres above sea level.

(16)

16

The study period was defined as the years of 2018 to 2019. The reason for not including previous years being because of the unusually high temperatures of 2018 in Sweden. 2018 and 2019 had the highest median temperature in Gothenburg in recent years and are therefore comparable to each other regarding summer temperatures, rather than if the study period had been extended with previous years. By doing so the risk of skewing the data in the analysis was reduced. Another reason for the decision to study the latest two years rather than previous years was because a similar study already had been conducted using much of the same data for the period of 2012 to 2017. (18)

When performing analyses, only the months with the highest temperatures were included, being the summer seasons of May to August each year. Swedish holidays were also controlled for using a binary indicator variable, as these could have a confounding effect on the results.

While the main focus of the study being psychiatric emergency visits in relation to high ambient temperatures, analyses were also performed on lower temperatures. Events for which analysis were performed and reported was for days when temperatures were above the 95th or below the 5th_percentile.

All psychiatric emergency visits in Gothenburg are accounted for, since the data is gathered from the only psychiatric emergency ward in the Gothenburg area. As such it is likely that the study yields sufficient statistical power to fairly accurately assess the risk ratio of the whole group. For subgroups of patients there might be some variation however regarding statistical power. Large groups such as males or females or patients aged 18-34 or 35-64 are likely to yield enough statistical power, while smaller groups of patients that make up a smaller percentage of the total number of visits are less likely to yield enough statistical power, such as patients with substance use disorders or patients over 65 years of age who less frequently seek psychiatric emergency care. Current and non-current outpatients were also fairly large groups of patients and should yield enough statistical power.

Statistical Methods

(17)

17

previous studies have shown that effects on health from temperatures often have a delayed effect. (14) The highest effect is often seen on the day of the event, (i.e. high temperature), but a delayed effect can be seen several days after the event. Since the previous study by Carlsen et. al (18) in the Gothenburg area used lag structures of 0-3 days and 0-14 days which yielded significant results, the same lag structures of 0-3 and 0-14 days was used in this study. Using this method, the effect of high temperatures at the day of the event and the cumulative effect during the following 3 or 14 days after the event are both accounted for.

The time stratified case-crossover design is a method where a day with a certain exposure, in this case temperature at a certain cut-off, is considered a case, and days who fulfil certain criteria, (same weekday, same month, same year), but do not have the same temperature cut-off are used as controls.

Since temperatures often display a great variation between days, weeks, months and years, and the association between temperature and visits was not expected to be linear, meaning the effect estimate would vary across the span of temperature, a non-linear model was used instead of a linear model.

In the modelling procedure, a crossbasis was specified, a function which combines two functions: one being the non-linearity of temperature values with a b-spline with two knots, and secondly allowing for different effects across lags. The crossbasis was used in a poisson regression model to estimate the effect of temperature on the number of visits across the time stratified strata. The model results were used to find the minimum effect temperature (the temperature value with the lowest number of outcomes), and then using the model results and the minimum effect temperature to predict the relative risk associated with temperature at certain temperature cut-offs (being the 95th and 5th percentile) compared with the minimum effect temperature for each outcome, including a 95% confidence interval.

(18)

18

Ethics

(19)

19

Results

Descriptive Data

A total of 246 days was included in this analysis, spanning the time periods of 1st of May to 31st of August of both 2018 and 2019.

A total of 9,142 psychiatric emergency visits occurred during this period, with a daily mean of 37.2 visits (standard deviation ±7.6). Minimum daily visits were 20, and maximum daily visits were 59. When performing analysis for current or non-current psychiatric outpatient, when only patients from the municipals of Gothenburg, Härryda, Partille and Mölndal were included, the number of visits was 7,286.Table 1 shows the total number of visits and daily averages for each patient category during the 246 days included in the analysis. The most common age category was 18-34 and there was a similar number of visits among men and women.

Table 1. Total number of visits and average daily visits for each patient category for the period of May-Aug 2018 and 2019.

Number of psychiatric emergency visits by category and daily averages

Total number of visits Daily visits average

Total visits 9142 37 Age 18-34 4364 18 Age 35-64 3861 16 Age 65+ 917 4 Male 4807 20 Female 4335 18 Current Outpatient 3031 12

(20)

20

Substance use disorder 2049 8

Temperature

For the months of May to August of 2018-2019, the temperature ranged between a minimum of 4.8°C and maximum of 27.4°C, with a mean temperature of 17.6°C. A graph depicting the daily temperature during the study period can be seen in Figure 1. The temperature percentiles were as follows: 1st percentile 6.3°C, 5th percentile 9.5°C, 95th percentile 23.7°C and 99th percentile 25.6°C. Results are reported for the 95th_{and 5}th_{percentile. These temperatures are}

shown in Table 2. Table 3 shows the minimum effect temperature, which was used as a reference point when calculating risk ratios for each group of patients.

Figure 1. Graph depicting daily temperature for the study period of May-Aug 2018 (black line) and May-Aug 2019 (red line).

(21)

21 Temperature data Mean Temperature 17.6°C Minimum Temperature 4.8°C Maximum Temperature 27.4°C 1st Percentile 6.3°C 5th Percentile 9.5°C 95th Percentile 23.7°C 99th Percentile 25.6°C

Table 3. Minimum effect temperature for lags 0-3 and 0-14 for each group of patients. Used to estimate risk ratios for each patient category.

Minimum Effect Temperature

Lags 0-3 Lags 0-14 Total 15.8°C 15.7°C Male 17.4°C 16.6°C Female 11.5°C 14.9°C Age 18-34 17.4°C 16.1°C Age 35-64 13.1°C 14.6°C Current outpatient 12.5°C 18.7°C

Not current outpatient 16.5°C 15.1°C

(22)

22

Risk ratios by patient category

Total

When analysing all patients, the risk ratio was significant for high temperatures both for lag periods of 0-3 and 0-14 days, being 12.0% and 20.6% respectively. There seems to also be an association between low temperatures below the 5th percentile and number of psychiatric emergency visits, being 11.1% and 47.4% respectively. However, for the lower temperatures these results were not statistically significant.

Figure 2. Total risk ratios for all patients to visit a psychiatric emergency ward in relation to temperature. Depicted for lags 0-3 and 0-14. Green line represents the estimated average risk ratio. Grey areas represent confidence intervals.

(23)

23

Males had a statistically significant association between extreme temperatures and psychiatric emergency visits, both at high and low temperatures. For high temperatures, these were 19.8% for 0-3 days lag periods, and 33.2% for 0-14 days lag periods, and for low temperatures the risk ratios were 29.6% for the 0-3 lag period and 61.3% for the 0-14 lag period, the former being statistically significant. The risk for males at both high and low temperatures was also numerically higher than the total risk ratio. While for females, the risk at the 95th or 5th was not statistically significant and the risk ratio for females was also numerically lower in comparison to the total risk ratio.

(24)

24

Figure 4. Risk ratios for females to visit a psychiatric emergency ward in relation to temperature. Depicted for lags 0-3 and 0-14. Green line represents the estimated average risk ratio. Grey areas represent confidence intervals.

Age

The younger age group, 18-34-year olds, showed significant results at high and low temperatures. For the 95th percentile, a risk ratio of 14.1% for 3 lag periods, and 27.3% for 0-14 lag periods was seen. For the 5th percentile, we saw an increase in risk ratio of 18.3% and 85.4% respectively for 0-3 and 0-14 lag periods, only the latter being statistically significant.

(25)

25

Because of the number of visits by those aged 65 years and older being far too few to have any statistical power in the age group spanning 65 years and older, we do not present results for this age group.

(26)

26

Figure 6. Risk ratios for patients aged 35-64 to visit a psychiatric emergency ward in relation to temperature. Depicted for lags 0-3 and 0-14. Green line represents the estimated average risk ratio. Grey areas represent confidence intervals.

Outpatients

(27)

27

(28)

28

Figure 8. Risk ratios for patients who are not a current psychiatric outpatient to visit a psychiatric emergency ward in relation to temperature. Depicted for lags 0-3 and 0-14. Green line represents the estimated average risk ratio. Grey areas represent confidence intervals.

Substance use disorder

(29)

29

Figure 9. Risk ratios for patients diagnosed with substance use disorder to visit a psychiatric

emergency ward in relation to temperature. Depicted for lags 0-3 and 0-14. Green line represents the estimated average risk ratio. Grey areas represent confidence intervals.

Table 4. Risk ratios of patients visiting a psychiatric emergency ward for lags 0-3 and 0-14 at temperatures of the 95th percentile (e.g. warmest days) and for the 5th percentile (e.g. coldest days). 95% Confidence intervals are shown within parenthesis.

Risk ratios at temperatures of the 95th_percentile

Lags 0-3 Lags 0-14

Total 1.120 (1.017-1.233)* 1.206 (1.060-1.372)*

(30)

30 Male 1.198 (1.052-1.364)* 1.332 (1.104-1.606)* Female 1.157 (0.954-1.403) 1.089 (0.895-1.325) Age categories Age 18-34 1.141 (1.018-1.279)* 1.273 (1.087-1.490)* Age 35-64 1.095 (0.926-1.295) 1.176 (0.943-1.466) Outpatient (yes/no) Current outpatient 1.049 (0.867-1.267) 1.083 (0.869-1.349)

Not current outpatient 1.137 (0.979-1.320) 1.199 (0.960-1.498)

Diagnosis

Substance use disorder 1.237 (0.985-1.555) 1.388 (0.139-13.778)

* = Significant results are marked with an asterisk.

Table 5. Risk ratios of patients visiting a psychiatric emergency ward for lags 0-3 and 0-14 at temperatures of the 5t percentile (e.g. coldest days). 95% Confidence intervals are shown within parenthesis.

Risk ratios at temperatures of the 5th percentile

(31)

31

Outpatient (yes/no)

Current outpatient 1.009 (0.895-1.137) 1.199 (0.628-2.292)

Not current outpatient 1.210 (0.948-1.544) 1.392 (0.730-2.654)

Diagnosis

Substance use disorder 1.344 (0.921-1.961) 1.62 .267-4.223)

* = Significant results are marked with an asterisk.

Discussion

The results further add to the evidence that there is a significant association between total risk ratio of psychiatric emergency visits and heat. This is in accordance with findings from the previous study conducted in the same area by Carlsen et al. (18), and by other studies conducted on the subject, see chapter “Emergency visits related to mental health during high ambient temperature”. Regarding specific demographic or diagnostic risk factors, increased risk ratios were seen for males and young adults at or above temperatures of the 95th_{percentile, and for}

the same groups during cold temperatures at certain lag periods. Surprisingly, the highest risk increase was seen at or below the 5th percentile for patients aged 18-34 at 85.4% (95% CI 1.122-3.066). This is not something that had been anticipated, as the relationship between cold temperatures during warm seasons has not been well studied previously. Interestingly, since the study period only included the summer period in Sweden of May to August, this effect was seen at temperatures that are otherwise around the average annual temperature.

Sex

(32)

32

statistically significant effect was not seen for the cumulative effect but was seen for each of the individual lag days of 2-6, the highest day being day 4 with a risk ratio of 1.616 (1.131-2.311). A statistically significant effect was also seen during lower temperatures between lag day 3-6 even though the increase in risk was lower than during high temperature, the highest risk ratio being 1.043 (1.013-1.074). Another study by Min et al. in Yancheng, China (27) conducted with similar methods to the current study, also found that males had an increased risk of visits during high temperatures at the 90th annual percentile, although results were just barely not significant.

The current study did not observe any risks for females. Although not significant, in comparison to the total risk ratio females even had a lower risk ratio overall during both warm and cold temperatures. According to this study then, it seems likely that females are not adversely affected by high or low temperatures. But from a study in Lisbon, by Almendra et al. (28), a significant association was seen for females at lags 0-4 during high temperatures, peaking at lag 2 at 1.36 (95% CI 1.08-1.70), but not any significant association for males. In this study, a protective effect was also seen for females at the 1st percentile. In contrast, Niu et al. found an increased risk for females at lower temperatures between lag day 4-6, peaking at 1.042 (1.007-1.079), but not for the cumulative effect. From these articles, females could have an increased risk at high temperatures, but contradicting results have been found for low temperatures.

Age

Young people did show a significant increase in risk ratio during heat, and during cold for lags 0-14. A study conducted by Min et al. also found a statistically significant increased risk ratio for those aged under 45 years during temperatures above the 90th annual percentile. They did not see a significant effect at lower temperatures however.

(33)

33

are at least one article that have shown significant correlation between cold and mental health, showing a risk ratio of 1.043 (95%CI: 1.017 -1.069) at or below the 10th annual percentile for all patients (26), but as far as the current author is aware, none that have reported such a huge effect from cold on the age group of 18-34, and none that have focused on the summer season alone.

The differences between age groups could possibly be explained by living habits. A reason might be that a large percent of the 18-34-year olds in Gothenburg are students and as such might have more free time during the summer to spend outside than those with a full-time job, and as such become more exposed to heat. Young people in Sweden might possibly also spend more time exposing themselves to the sun in hopes of getting tanned than those middle-aged causing further heat exposure, but that is only speculation.

Analysis were not performed for patients aged 65 years or older in the study because of the lack of statistical power caused by the low number of visits in that age group. That group should not be excluded however from further studies, as some studies have shown that the elderly with psychiatric disorders are indeed impacted by heat. The study from Hansen et al. in Adelaide, Australia (11) showed an increased mortality risk for the those with mental and behavioural disorders aged 65-74 during heatwaves. The study by Hansen measured all deaths of people with mental illness however, including deaths that are not directly affected by their mental health status. In another article by Liu et al., an increased odds ratio of 3.034 (95% CI 1.802-5.139) of patients aged 65 or older visiting an emergency ward for psychiatric care during heat waves was seen. (19). It is very likely that there is an effect of heat on mental health in the elderly, even though it could not be studied in this research.

Outpatients and non-outpatients

(34)

34

total risk ratio, most likely indicating that outpatients are not affected by temperature, or perhaps only by a small margin. One could argue that a reason for this could be that outpatients already have an ongoing contact with a specialized psychiatric practitioner or unit, and in case of sudden exacerbation are more likely to contact that unit instead of seeking emergency care. There is also a possibility that their ongoing treatment is working and has a protective effect on the outpatient, meaning a possible exacerbation by heat or cold is prohibited by their ongoing treatment. It should also be noted that all outpatients include a broad range of diagnoses, and that patients with a specific diagnosis might be more affected by heat than others. If that diagnosis is rare however, those patients are less likely to have an impact on the risk of all outpatients, as measured in this study. Therefore, it might still be of value in future research to study specific psychiatric diagnoses in relation to temperature. The overall increase in psychiatric emergency visits during heat are either way very unlikely to be caused by psychiatric outpatients, and strongly indicate that the increase of total visits cannot be explained by chronic psychiatric illness alone.

Substance use disorder

(35)

35

Another thing to note is that the diagnosis set at the psychiatric emergency ward is not always the most accurate diagnosis. While a diagnosis should not be set that is incorrect, it is possible that the emergency ward underdiagnoses patients or records a broad diagnosis since time in the emergency ward is limited, and further tests and time are often needed before a more specific diagnosis can be determined. In this study however this only applies to substance use disorders, as this was the only diagnosis included in the study.

Another aspect to keep in mind is that all patients diagnosed with substance use disorder (ICD-codes: F10-19 or F55) was included, both those with one of these diagnosis as a main or additional diagnosis at the time of the visit. Because of this, patients with previous substance use disorder that are seeking psychiatric emergency care for other reasons than their substance use could also be included in this patient group. In these cases, however, it is possible that the underlying substance use could still be a determining factor in the exacerbation or triggering of other mental illnesses, and that heat plays an important role in triggering such exacerbations. It was therefore decided to include all cases of substance use disorder, not only when defined as a main diagnosis.

Strengths and limitations

A timeseries-based analysis was used in this study, using the distributed lag non-linear model, which takes into account the lagged effects of temperature (14), and also accounts for variations in daily and monthly visits. Lately several other articles have been published that a similar methods, and there seems to be lots of evidence that the effect of heat on mental health are distributed over time over several days. (17,19,26,27). By doing so this study considers effects that might otherwise be overlooked if only the days of exposure had been analysed.

(36)

36

emergency care in Sweden is rarely limited by socioeconomic factors, as the cost for medical or psychiatric care in Sweden is generally very low. This ensures that the patients in our study could come from a wide range of socioeconomic backgrounds. The only exclusion criteria used was for current or non-current outpatients, where patients were excluded that did not live in one of three of the neighbouring municipals of Gothenburg. The reason for this being that outpatient data was limited to patients within these areas.

In this study, the only diagnosis measured was substance use disorders. The reason for this being that when stratifying by diagnosis, substance use disorders were the only diagnosis which had enough visits to be likely to achieve statistical power. Ideally, other diagnosis could have been included too.

It should be kept in mind that by using the method of studying the number psychiatric emergency visits, all cases of mental illness might not be included. Patients might seek help at primary care centres instead of emergency wards, and outpatients might be able to directly contact their psychiatrist or care unit. Mortalities related to mental health are also not fully included by this method, neither are suicides. Keeping this in mind, the actual total effect heat has on mental health is not able to be measured simply by analysing psychiatric emergency visits. The combination of different methods is needed to fully understand the broad scope of how mental illnesses can be impacted by high or low temperatures.

2018 was an unusually hot summer in Sweden. Even though the average temperature was higher than previous years, especially during the month of May, an effect of heat could still be seen at or above the 95th percentile on total visits. There are studies that indicate that the effect of heat on general health is not determined by the temperature baseline, but rather by sudden changes in temperature (20), which is in accordance to the current study as the summers of 2018 and 2019 had a higher median temperature than previous years but still showed significant results at the 95th percentile.

(37)

37

not skew the data during analysis. By doing so the analysis lost some of its intended power, but there were still several statistically significant results to be seen.

A weakness in this study is that the temperature of the metrological measuring station is not a completely accurate measure of the surrounding temperature of each patient. Temperatures might vary slightly across the geographical area where temperature was measured, and socioeconomic factors for example might play a part in how well an individual can adapt to heat (by the presence of air condition in the living quarters for example), which was not controlled for in this study. It would however be very difficult to monitor surrounding temperatures for each patient in a large-scale study.

One patient was not limited to a single visit during the study period, meaning that if one patient visited the emergency ward several times, every visit was included in the analysis. This could have some impact on the results. However, since the analysis are performed by comparing number of visits between hot or cold periods to periods with normal temperatures, such multiple visits from the same patient are likely to balance each other out, at least if those visits are independent of ambient temperatures. If they are dependent of ambient temperatures, there is a possibility of over- or underestimating the risk ratio if a small number of patients seek psychiatric emergency care very often at certain temperatures. But there is no objective approach to limit each patient to a single visit in the study. If only one visit would count per patient, an arbitrary choice would have to be made whether to include a visit during extreme temperatures or normal temperatures, making the results of the study questionable.

(38)

38

Another factor which was not accounted for in this study could be the use of medications, such as anticholinergic drugs that could cause hypohidrosis and thereby a lack of ability to regulate body heat. There could also be other somatic illnesses that could predispose a patient to be more susceptible to heat, which might in turn cause them to experience a worsening of their mental health as well. These confounding factors could cause the risk to be incorrectly estimated if a fair portion of patients were affected by them.

Another confounding factor is the daily, weekly and monthly variation in psychiatric emergency visits. But by using a timeseries model this variation is accounted for, as the number of visits during a certain day of the week is only compared to the same weekdays within the same month and year. As such, the natural variation of visits between weekdays, months and year that can be observed are adjusted for when estimating risk ratios. (The observation of the data showed that there are on average more psychiatric emergency visits on a Monday for example than on a Saturday).

While this and other studies have shown a strong association between heat and the number psychiatric emergency visits, there is not for certain that there is a causal relationship between them. While the association might be caused by physiological reactions to heat, there might be other reasons than heat itself that causes an increased number of people seeking psychiatric emergency care. During the summer holidays or vacation for example, people living alone might feel more isolated than the time when they’re working or studying, causing their mental health to worsen. However, this should not be seen only during the warmest summer days in such a case but rather is likely to be seen over the whole summer period. But there might be other factors which are only indirectly related to heat that could cause people to seek psychiatric care during the warmest days, for example air humidity or sun exposure, which are likely to increase during warm days.

(39)

39

and probably unrealistic at present to seek to protect such large groups from heat exposure. To be able to protect certain groups of vulnerable patients, further research is needed to determine more specific risk factors before any protective measures could be carried out.

Conclusions

The aim of the study was to assess how mental health in certain groups of patients were influenced by heat and cold by analysing the number of psychiatric emergency visits during warm or cold days. Overall, the conclusion is that heat influence the total amount of visits to the psychiatric emergency ward in Gothenburg. Male and young adults aged 18-34 had an increased risk ratio of visiting a psychiatric emergency ward during warm days, and by association more likely to experience a worsening of mental health during high temperatures. During cold temperatures, males had a higher risk of seeking psychiatric emergency care, as well as young adults aged 18-34 who had a high increased risk during 0-14 lag-periods at cold temperatures. There was also a non-significant increase in risk observed for patients with substance use disorders at both high and low temperatures. For outpatients there was a trend pointing towards a lower risk than the total risk for both high and low temperatures, strongly indicating that outpatients are not a major contributor to the total amount of patients visiting the psychiatric emergency ward.

(40)

40

Populärvetenskaplig sammanfattning

Projekttitel: Grupper i befolkningen som löper ökad risk att söka psykiatrisk akutvård efter att ha varit utsatta för höga eller låga utomhustemperaturer.

Värme påverkar många negativt, särskilt de som är äldre eller de som lider av hjärt- eller njursjukdom. Vad många inte vet är att på senare tid har forskning visat att värme verkar ha en negativ effekt även på vår psykiska hälsa. Flera studier världen över har visat, bland annat i Göteborg, att antalet personer som söker psykiatrisk akutsjukvård ökar när det är som varmast utomhus.

I den här studien var målet att undersöka detta fenomen vidare, och ta reda på om det finns specifika grupper som är särskilt utsatta för värme. Detta gjordes genom att titta på vilka personer som sökte psykiatrisk akutsjukvård när det var som varmast respektive kallast under sommaren 2018 och 2019 och jämföra det med antalet besök under mer normala temperaturer.

Totalt under de åtta månader (fyra per år) som studien tittade på skedde 9142 besök till Östra sjukhusets psykiatriska akutmottagning.

När man tittade på specifika grupper av patienter sågs det att män hade en ökad risk att söka psykiatrisk akutvård vid höga utomhustemperaturer. Risken ökade för män med upp till 33,2% jämfört med normala temperaturer. Unga vuxna, alltså de i åldern 18-34 år, hade också en ökad risk att söka psykiatrisk akutvård vid värme, en risk som uppgick till 27,3% jämfört med normala temperaturer.

Vid kyla upptäcktes också en ökad risk för både män och unga vuxna. Män hade en ökad risk på 29,6% att söka psykiatrisk akutvård, och unga vuxna en ökad risk på hela 85,4%, i jämförelse med normala temperaturer.

(41)

41

I framtiden kan man förhoppningsvis forska på vad exakt det är som gör att män och unga vuxna är mer utsatta för dessa temperaturer, och om det finns andra grupper i befolkning som också har en ökad risk. Om man kan komma att förstå varför så skulle det i framtiden vara möjligt att kunna ge en bättre vård till dessa personer, och även att arbeta förebyggande för att skydda dessa personer från värmens eller kylans skadliga effekter på mental hälsa.

Acknowledgements

A Special thanks to Daniel Oudin who helped with the analysis, and to Peter Asplund who helped extract the patient data used in this study. I also want to extend a very special thanks to my supervisors Steinn Steingrimsson and Hanne Krage Carlsen who helped me during this project.

Reference list

1. Dematte JE, O’Mara K, Buescher J, Whitney CG, Forsythe S, McNamee T, m.fl. Near-fatal heat stroke during the 1995 heat wave in Chicago. Ann Intern Med. 01 augusti 1998;129(3):173–81.

2. Robine J-M, Cheung SLK, Le Roy S, Van Oyen H, Griffiths C, Michel J-P, m.fl. Death toll exceeded 70,000 in Europe during the summer of 2003. Comptes Rendus Biologies. 01 februari 2008;331(2):171–8.

3. Perkins SE, Alexander LV, Nairn JR. Increasing frequency, intensity and duration of observed global heatwaves and warm spells. Geophysical Research Letters [Internet]. 2012 [citerad 18 maj 2020];39(20). Tillgänglig vid: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2012GL053361

4. Meehl GA, Tebaldi C, Tilmes S, Lamarque J-F, Bates S, Pendergrass A, m.fl. Future heat waves and surface ozone. Environ Res Lett. maj 2018;13(6):064004.

5. Nitschke M, Tucker GR, Bi P. Morbidity and mortality during heatwaves in metropolitan Adelaide. Medical Journal of Australia. 2007;187(11–12):662–5.

6. Durkheim É. Suicide: a study in sociology. Glencoe, Illinois: Free Press; 1951.

(42)

42

8. Basu R, Gavin L, Pearson D, Ebisu K, Malig B. Examining the Association Between Apparent Temperature and Mental Health-Related Emergency Room Visits in California. Am J Epidemiol. 01 2018;187(4):726–35.

9. Kim Y, Kim H, Gasparrini A, Armstrong B, Honda Y, Chung Y, m.fl. Suicide and Ambient Temperature: A Multi-Country Multi-City Study. Environ Health Perspect. 2019;127(11):117007.

10. Burke M, González F, Baylis P, Heft-Neal S, Baysan C, Basu S, m.fl. Higher temperatures increase suicide rates in the United States and Mexico. Nature Climate Change. augusti 2018;8(8):723–9.

11. Hansen A, Bi P, Nitschke M, Ryan P, Pisaniello D, Tucker G. The Effect of Heat Waves on Mental Health in a Temperate Australian City. Environ Health Perspect. oktober 2008;116(10):1369–75.

12. Page LA, Hajat S, Kovats RS, Howard LM. Temperature-related deaths in people with psychosis, dementia and substance misuse. The British Journal of Psychiatry. juni 2012;200(6):485–90.

13. Obradovich N, Migliorini R, Paulus MP, Rahwan I. Empirical evidence of mental health risks posed by climate change. PNAS. 23 oktober 2018;115(43):10953–8.

14. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med. 20 september 2010;29(21):2224–34.

15. Ye X, Wolff R, Yu W, Vaneckova P, Pan X, Tong S. Ambient Temperature and Morbidity: A Review of Epidemiological Evidence. Environmental Health Perspectives. januari 2012;120(1):19–28.

16. Muggeo VM, Hajat S. Modelling the non-linear multiple-lag effects of ambient temperature on mortality in Santiago and Palermo: a constrained segmented distributed lag approach. Occup Environ Med. september 2009;66(9):584–91.

17. Wang X, Lavigne E, Ouellette-kuntz H, Chen BE. Acute impacts of extreme temperature exposure on emergency room admissions related to mental and behavior disorders in Toronto, Canada. Journal of Affective Disorders. 01 februari 2014;155:154–61.

18. Carlsen HK, Oudin A, Steingrimsson S, Oudin Åström D. Ambient Temperature and Associations with Daily Visits to a Psychiatric Emergency Unit in Sweden. Int J Environ Res Public Health. 21 2019;16(2).

(43)

43

[Internet]. januari 2019 [citerad 09 mars 2020];16(1). Tillgänglig vid: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6339177/

20. Nordio F, Zanobetti A, Colicino E, Kloog I, Schwartz J. Changing patterns of the temperature-mortality association by time and location in the US, and implications for climate change. Environ Int. augusti 2015;81:80–6.

21. Bobb JF, Peng RD, Bell ML, Dominici F. Heat-related mortality and adaptation to heat in the United States. Environ Health Perspect. augusti 2014;122(8):811–6.

22. Åström DO, Forsberg B, Edvinsson S, Rocklöv J. Acute Fatal Effects of Short-Lasting Extreme Temperatures in Stockholm, Sweden: Evidence Across a Century of Change. Epidemiology. november 2013;24(6):820–9.

23. Relationship Between Ambient Temperature and Humidity and Visits to Mental Health Emergency Departments in Québec | Psychiatric Services [Internet]. [citerad 24 mars 2020]. Tillgänglig vid: https://ps-psychiatryonline-org.ezproxy.ub.gu.se/doi/full/10.1176/appi.ps.201100485

24. Lee S, Lee H, Myung W, Kim EJ, Kim H. Mental disease-related emergency admissions attributable to hot temperatures. Science of The Total Environment. 01 mars 2018;616– 617:688–94.

25. Vida S, Durocher M, Ouarda TBMJ, Gosselin P. Relationship Between Ambient Temperature and Humidity and Visits to Mental Health Emergency Departments in Québec. PS. 01 november 2012;63(11):1150–3.

26. Niu Y, Gao Y, Yang J, Qi L, Xue T, Guo M, m.fl. Short-term effect of apparent temperature on daily emergency visits for mental and behavioral disorders in Beijing, China: A time-series study. Science of The Total Environment. 28 april 2020;139040.

27. Min M, Shi T, Ye P, Wang Y, Yao Z, Tian S, m.fl. Effect of apparent temperature on daily emergency admissions for mental and behavioral disorders in Yancheng, China: a time-series study. Environ Health. 20 2019;18(1):98.

28. Almendra R, Loureiro A, Silva G, Vasconcelos J, Santana P. Short-term impacts of air temperature on hospitalizations for mental disorders in Lisbon. Sci Total Environ. 10 januari 2019;647:127–33.

(44)

44

Demographic risk factors increasing risk of seeking psychiatric emergency care after heat exposure

SAHLGRENSKA ACADEMY