Article Text

Download PDFPDF
Characteristics and outcomes of accidental hypothermia in Japan: the J-Point registry
  1. Tasuku Matsuyama1,
  2. Sachiko Morita2,
  3. Naoki Ehara3,
  4. Nobuhiro Miyamae4,
  5. Yohei Okada5,
  6. Takaaki Jo6,
  7. Yasuyuki Sumida7,
  8. Nobunaga Okada1,8,
  9. Makoto Watanabe1,
  10. Masahiro Nozawa9,
  11. Ayumu Tsuruoka10,11,
  12. Yoshihiro Fujimoto12,
  13. Yoshiki Okumura13,
  14. Tetsuhisa Kitamura14,
  15. Bon Ohta1
  1. 1 Department of Emergency Medicine, Kyoto Prefectural University of Medicine, Kyoto, Japan
  2. 2 Department of Emergency and Critical Care Medicine, Senri Critical Care Medical Center, Saiseikai Senri Hospital, Suita, Japan
  3. 3 Department of Emergency Medicine, Japanese Red Cross Kyoto Daiichi Hospital, Kyoto, Japan
  4. 4 Department of Emergency Medicine, Rakuwa-kai Otowa Hospital, Kyoto, Japan
  5. 5 Department of Emergency and Critical Care Medicine, Japanese Red Cross Society Kyoto Daini Red Cross Hospital, Kyoto, Japan
  6. 6 Department of Emergency Medicine, Uji-Tokushukai Medical Center, Uji, Japan
  7. 7 Department of Emergency Medicine, North Medical Center, Kyoto Prefectural University of Medicine, Yosa-gun, Japan
  8. 8 Department of Emergency and Critical Care Medicine, National Hospital Organization, Kyoto Medical Center, Kyoto, Japan
  9. 9 Department of Emergency and Critical Care Medicine, Saiseikai Shiga Hospital, Ritto, Japan
  10. 10 Department of Emergency and Critical Care Medicine, Kidney and Cardiovascular Center, Kyoto Min-iren Chuo Hospital, Kyoto, Japan
  11. 11 Department of Emergency and Critical Care Medicine, Emergency and Critical Care Medical Center, Osaka City General Hospital, Osaka, Japan
  12. 12 Department of Emergency Medicine, Yodogawa Christian Hospital, Osaka, Japan
  13. 13 Department of Emergency Medicine, Fukuchiyama City Hospital, Fukuchiyama, Japan
  14. 14 Department of Social and Environmental Medicine, Division of Environmental Medicine and Population Sciences, Graduate School of Medicine, Osaka University, Suita City, Japan
  1. Correspondence to Dr Tasuku Matsuyama, Department of Emergency Medicine, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan; task-m{at}


Background Accidental hypothermia (AH) has higher incidence and mortality in geriatric populations. Japan has a rapidly ageing population, and little is known about the epidemiology of hypothermia in this country.

Methods We created an AH registry based on retrospective review of patients visiting the ED of 12 institutions with temperature ≤35°C between April 2011 and March 2016. The severity of AH was classified as mild (≤35, ≥32°C), moderate (<32, ≥28°C) or severe (<28°C). The relationship between in-hospital mortality and severity of AH was assessed using a multivariable logistic regression analysis.

Results A total of 572 patients were registered in this registry and 537 patients were eligible for our analysis. The median age was 79 (IQR 66–87) years and the proportion of men was 51.2% (273/537). AH was more likely to occur in elderly patients aged ≥65 years (424/537, 80.0%) and in indoor settings (418/537, 77.8%). The condition most frequently associated with AH, irrespective of severity, was acute medical illness. A lower mean outside temperature was associated with a higher prevalence of AH, and particularly severe AH (p for trend <0.001). The overall proportion of cases resulting in in-hospital death was 24.4% (131/537), with no significant difference between severity levels observed in a multivariable logistic regression analysis (severe group (37/118, 31.4%) vs mild group (42/192, 21.9%), adjusted OR (AOR) 1.01, 95% CI 0.61 to 1.68; and moderate group (52/227, 22.9%) vs mild group, AOR 1.11, 95% CI 0.58 to 2.14).

Conclusion Active prevention and intervention should occur for this important public health issue.

  • environmental medicine
  • epidemiology
  • geriatrics

Statistics from

Key messages

What is already known on this subject

  • Accidental hypothermia is linked to high mortality and morbidity.

  • Furthermore, accidental hypothermia has higher incidence and mortality in geriatric populations.

What this study adds

  • This multicentre study indicated that elderly patients aged ≥65 years and indoor occurrence of accidental hypothermia accounted for approximately 80% of our registry.

  • One-quarter of accidental hypothermia cases were associated with death, although no significant difference in mortality was observed based on the severity of the accidental hypothermia.

  • Our findings offer important clues for both the prevention of accidental hypothermia and appropriate interventions.


Accidental hypothermia (AH) is defined as an involuntary drop in body core temperature to a level ≤35°C. AH is linked to high mortality and morbidity.1 2 According to a study in the USA, in-hospital mortality for patients with moderate-to-severe AH was approximately 40%, while an earlier report from Japan found that about 30% of all ED presentations for AH resulted in death.3 4 Importantly, geriatric populations were reported to be at an increased risk of AH and subsequent death.5 6 Therefore, AH will pose an increasingly heavy burden on industrialised countries, including Japan, with the rapid ageing of the population.

While a higher incidence of AH is documented in regions with severe winters, it also occurs in milder climates.7–9 In Japan, with its milder climate, most studies on AH were performed at a single centre, and a large multicentre questionnaire only focused on 3 months during winter.4 10 11 Thus, little epidemiological information about AH in Japan is available, which is a barrier to developing effective prevention and intervention strategies. Furthermore, uncertainty exists about the optimal management of patients with AH. There are a wide variety of recommended rewarming procedures,12 but no one procedure offers an advantage over the others in terms of improved outcomes, particularly for patients without cardiac arrest. For patients with both AH and cardiac arrest, studies are biased towards successfully resuscitated cases, and there is not enough evidence to guide the most appropriate management and resource allocation.13 14

In this multicentre observational study, our objects were to (1) evaluate the prevalence of AH and its association with characteristics such as social factors and climates, (2) describe the actual management for AH according to its severity and (3) investigate outcomes for patients with AH.


Study design and setting

We conducted the Japanese accidental hypothermia network registry (J-Point registry) based on retrospective review of patients visiting the ED of 12 institutions with temperature ≤35°C to obtain epidemiological information on AH, and to characterise its management in Japanese EDs, with the aim of improving outcomes. The Japanese AH network consists of eight critical care medical centres (CCMCs) and four non-CCMCs with an ED across the Kyoto, Osaka and Shiga Prefectures in Japan. Participating institutions had a median annual ED visit volume of 19 651 (IQR 13 281–27 554).

Study subjects

We retrospectively identified eligible patients using the International Classification of Diseases, Tenth Revision (ICD-10) code T68: ‘Hypothermia’, who were diagnosed during the study period, 1 April 2011 to 31 March 2016. We excluded those who aged <18 years, did not visit a participating ED, whose body temperature was unknown or >35°C, or where the patient or their family members had refused to be part of the registry. The Ethics Committee of each institution approved this study protocol. This study did not need to obtain each patient’s or their family member’s informed consent, but the Ethics Committee of each institution requested that we should use ‘opt-out consent’. Therefore, the guidance on opt-out consent was posted on the board of emergency room or the website of each participating institution.

Data collection and quality control

Data for the J-point registry were collected using a predefined uniform data sheet. All chart reviewers were emergency physicians who were trained in appropriate data extraction during face-to-face or web meetings. The chart reviewer at each participating institution performed a preliminary registration of two or three eligible patients’ data to check the feasibility of the data sheet. Several modifications were then made to the data sheet, and final data extractions were completed. The collected data were checked by the J-Point registry working group and confirmed or returned to the institution for clarification if an issue was identified, before being entered into the registry.

Patient-level measurements

The following baseline patient information was collected: sex, age, activities of daily living (ADL) before the AH (independent, needing some assistance, needing total assistance), residence (living at home alone, living at home not alone, nursing home, homeless), medical history (cardiovascular disease (ischaemic heart disease, heart failure, arrhythmia, hypertension, other), neurological disease (stroke, epilepsy, Parkinson disease or syndrome, other), endocrine disease (diabetic mellitus, thyroid diseases, adrenal insufficiency, other), psychiatric disease (chronic alcoholic, depression, schizophrenia, other), malignant disease, dementia, other), mean outdoor temperature on the day AH occurred, season, location (indoor, outdoor) and mode of arrival (walk-in, ambulance).

In-hospital level measurements

Collected in-hospital data included vital signs on arrival at hospital (body temperature, blood pressure, heart rate, GCS score), biological data (serum pH, HCO3 (mEq/L), lactate (mmol/L), sodium (mEq/L), potassium (mEq/L), glucose (mg/dL)), cold exposure, associated conditions, treatment process and outcome. A definition of cold exposure was clinical judgement by clinician who cared for a patient or who entered the data of this study. Sequential organ failure assessment (SOFA) score was only calculated for patients admitted to the intensive care unit (ICU).

In line with a previous study,4 associated conditions were classified into acute medical illness (stroke, seizure, Parkinson disease, thyroid disease, hypoglycaemia, infectious disease, acute pancreatitis, uraemia, malignant disease, bowel ischaemia, rhabdomyolysis, other), traumatic injury (fall (head, spine, extremity injury), motor vehicle accident, other), alcohol intoxication, drowning (indoor, outdoor), self-harm (drug, external) and other (iatrogenic, mountain incident, burn, malnutrition/infirmity, other). The diagnoses of acute medical illness were obtained from ICD-9 or 10 code in the final medical summary or medical records. Rewarming procedures were divided into active external/minimally invasive rewarming (warm intravenous fluids, warm blanket, forced warm air, heating pads, warm bath) and active internal rewarming (lavage (stomach, chest, bladder), intravascular haemodialysis, extracorporeal membrane oxygenation).15 Other treatment information included endotracheal intubation, use of catecholamines and emergent transvenous cardiac pacing. Data collected regarding outcomes were in-hospital death, major complications during rewarming such as ventricular fibrillation (VF) or pulseless ventricular tachycardia (VT), and length of hospital stay.

Outcome measures

The primary outcome of this study was in-hospital death. Secondary outcomes were cardiac arrests and VF or pulseless VT during rewarming, and length of hospital stay in survivors.

Statistical analysis

Based on the common classification of the severity of AH,1 2 patients were divided into three groups: mild (32°C–35°C), moderate (28°C–31°C) or severe (<28°C). Patient characteristics, in-hospital information and outcomes for the three severity groups were compared using Kruskal-Wallis tests for continuous variables and χ2 tests or Fisher’s exact tests for categorical variables. The association between each variable considered to be potentially associated with clinical outcomes and in-hospital was examined with the use of χ2 tests. We described the prevalence of AH according to climate information such as the season and mean outside temperature on the day that the patient visited the ED. Mean outside temperatures were retrieved from the Japan Meteorological Agency website.16

The association between severity categories and in-hospital death was investigated through two models of multivariable logistic regression analyses, with adjusted ORs (AORs) and their 95% CI as the effect variables. For the multivariable analysis in model 1, we selected potential confounders considered to be associated with clinical outcomes including age categories (18–64 years, 65–74 years, ≥75 years), sex (men, women), ADL (independent, not independent, unknown), systolic blood pressure (cardiac arrest, unmeasurable, 40–90 mm Hg, <90 mm Hg), presence of associated acute medical illness (yes, no), active internal rewarming (yes, no) and hospital type (CCMC, non-CCMC). In model 2, we selected the factor of ‘the number of comorbidities identified in the medical history (none, one, multiple, unknown)’ in addition to the factors included in model 1. We treated the missing data of each variable as ‘an unknown category’. All p values were two-sided and levels <0.05 were considered statistically significant. All statistical analyses were carried out using SPSS software (V.24J).


During the study period, a total of 572 patients were entered into the J-point registry. There were no multiple hypothermia events from one patient and no patient was excluded from this study for the reason that the patients or their family members refused to be a part of this registry. After excluding 24 patients whose body temperature was >35°C, 3 patients with an unknown body temperature and 8 patients aged <18 years, we analysed 537 patients with AH. There were 192, 227 and 118 patients in the mild, moderate and severe groups, respectively.

Patient characteristics and associated climatic conditions

About half the patients in each group were men. The proportion of patients aged ≥65 years was 80.0% (424/537), and over 60% of each group was ≥75 years. Approximately 70% of patients were independent in ADL. Most patients resided at home (479/537, 89.2%), with the highest proportion of home dwellers in the severe group (111/118, 94.1%). AH was more likely to occur in indoor settings (425/537, 77.8%), while the mild group experienced the highest proportion of AH in outdoor settings (59/192, 30.7%) (table 1). The prevalence of AH was highest in winter (351/537, 65.4%) and lowest in summer (19/537, 3.5%) (figure 1). A low mean outside temperature was associated with a higher prevalence of AH, particularly in the severe group (p for trend <0.001) (figure 2).

Figure 1

Prevalence of accidental hypothermia by month and severity.

Figure 2

Prevalence of accidental hypothermia by severity and mean outside temperature on day of the arrival at the ED. A higher proportion of severe hypothermia occurred with lower mean outside temperatures (p for trend <0.001).

Table 1

Patient characteristics according to the severity of accidental hypothermia

In-hospital data

Median body temperature was 30.8°C (IQR 28.2°C–32.6°C). Members of the severe group were more likely to have unstable vital signs and a higher SOFA score (all p<0.001). The proportion of cold exposure was highest in the severe group, followed by the moderate and mild groups (p<0.001). The condition most frequently associated with AH, irrespective of severity, was acute medical illness. More than 90% of each group received active external/minimally invasive rewarming procedures. The severe group was most likely to be rewarmed by active internal methods and experience endotracheal intubation, catecholamines and ICU admission (table 2).

Table 2

In-hospital data


The overall proportion of in-hospital deaths was 24.4% (131/537), with no significant difference between the severity groups (21.9% (42/192), 22.9% (52/227) and 31.4% (37/118) in the mild, moderate and severe groups, respectively, p=0.133) (table 3 and online supplementary table 1). This lack of a significant difference between severity groups remained after multivariable logistic regression analyses (severe vs mild group, AOR 1.01, 95% CI 0.61 to 1.67 in model 1 and AOR 1.02, 95% CI 0.62 to 1.70 in model 2; moderate vs mild group, AOR 1.04, 95% CI 0.54 to 1.99 in model 1 and AOR 1.09, 95% CI 0.56 to 2.10 in model 2). However, in-hospital deaths were independently associated with sex, age, blood pressure and associated acute medical illness among patients, and also the type of hospital in which patient received treatment (table 4).

Supplemental material

Table 3

Outcomes according to the severity of accidental hypothermia

Table 4

Factors associated with in-hospital death

With regard to secondary outcomes, the severe hypothermia group experienced a higher proportion of cardiac arrests, VF or pulseless VT during rewarming (each p=0.032 and 0.001) and also had a significantly longer length of hospital stay for survivors (p<0.001) (table 3).


Data in the registry developed as part of this multicentre study demonstrated that elderly patients aged ≥65 years and indoor occurrence accounted for approximately 80% of cases of AH. There was a strong association between AH and acute medical illness, and the prevalence of severe cases of AH was higher in colder climates. Even after adjusting for potential confounders, there was no significant relationship between severity of AH and in-hospital mortality. To the best of our knowledge, this was the first multicentre study to assess the epidemiology of AH in Japan over several years and one of the largest descriptive studies of AH in the world. Our findings offer important clues for both the prevention of AH and appropriate interventions.

Our results demonstrated no significant difference in the in-hospital mortality among the three groups of the severity. Our findings were likely multifactorial. First, vital signs such as systolic blood pressure had a significant association with higher in-hospital mortality. Indeed, vital signs were worse in the severe group, and this group had a higher likelihood of receiving intensive treatment such as active internal rewarming procedures, endotracheal intubation or the use of catecholamines. Our results demonstrated that among patients with unstable vital signs, haemodynamic status had a greater impact on the outcome than the severity of the hypothermia itself. Second, unlike other studies that found the predominant condition associated with secondary hypothermia to be alcohol or drug intoxication,3 6 about half of the patients in each severity group in this study had acute medical illness, and these underlying conditions should be treated in addition to rewarming. Importantly, patients with secondary hypothermia often died due to an underlying disease rather than the hypothermia itself.15 17 Indeed, our findings observed that the presence of an associated medical condition was significantly related to the higher prevalence of in-hospital death. In this study, environmental factors had less influence on the mild group, which suggests that underlying conditions may themselves have been the cause of hypothermia and have a higher likelihood of resulting in in-hospital death. Third, in this study with the extremely old age population, when taking the effect of patient’s age and decreased ADL into consideration, such basic factors were more important than body temperature in order to overcome this high mortality condition. Therefore, for this population, current AH classification might reflect the degree of the patient’s decreased body temperature, but not the severity of their overall condition. Finally, this study observed that patients with AH treated in CCMCs were associated with decreased in-hospital mortality. Previous studies also demonstrated that in various acute critical illnesses, patients treated in specialised centres such as tertiary centres or CCMCs had a positive association with better outcome.18 19 These results might be applicable to patients with AH.

Our findings identified that most cases of AH occurred in winter and that colder climates were associated with larger numbers of severe cases. These results were largely consistent with previous reports from other western countries, although mean outside temperatures showed substantial variation based on the climate of the study settings.6 7 9 Although the actual incidence and mortality rates of AH were unknown in Japan, cold exposure was the leading cause of weather-related deaths. Similarly, in the USA, approximately twice as many deaths were attributed to cold exposure as to heat exposure.20 Despite the heavy disease burden, less attention is paid to AH than to heatstroke in Japan’s public health programme. Japan has made a great effort to increase public awareness around prevention and intervention for heatstroke by updating the risk of heatstroke every day based on the Wet-Bulb Globe Temperature.21 Greater awareness, public education and effective prevention strategies and interventions for patients with AH should be developed as nationwide initiatives in Japan.

We found that about 80% of patients with AH were aged ≥65 years, and of these about three-quarters were aged >75 years. Although several studies have highlighted a higher prevalence of AH in the elderly,3 6 22 the proportion of elderly was even higher in our study. Diminished thermoregulatory functions in the elderly limit their tolerance to changes in temperature.23 They may also fail to initiate compensatory responses to a decreasing body temperature, such as wearing suitable clothing or controlling the air conditioning, because of dementia or impaired consciousness due to a decreased body temperature. Importantly, the number of elderly people in the population continues to grow rapidly worldwide. For example, the proportion of the Japanese population aged ≥75 years is estimated to increase from 7.1% in 2000 to 18.1% by 2025.24 Developing an effective system for managing AH in the elderly is an urgent public health issue for Japan and other industrialised countries.

In this study, most of the patients lived at home, and AH occurred most frequently in indoor settings. Social isolation, such as homelessness or living alone, was reported to be a key risk factor among patients with AH.5 The proportion of homeless patients in our registry was 1.7%, which was low compared with a previous report from the USA.5 The number of homeless in Japan has decreased since the initiation of an active intervention for them in 2002.25 However, the proportion of patients in our registry who lived at home alone was approximately 40%. According to a report from the Japanese Consumer Affairs Agency, the elderly, even when living at home with others, were more likely to be isolated, as they were less likely to leave the house and spent 42.3% of their waking hours alone.26 As the population ages, the number of socially isolated people will increase.26 Therefore, active interventions such as developing communication networks and improving the residential environment are also needed.


Several limitations have to be considered in this study. First, this was a retrospective chart review, so we could not avoid the presence of missing data. However, most variables, other than the GCS score and biological data, had less than 5% missing data and did not affect our results significantly. In addition, most variables in this study could be evaluated objectively, so subjectiveness and bias on the part of the data extractor were minimised. Second, we selected study subjects based on the ICD 10 code and may have missed patients whose body temperature was ≤35°C. Third, this study may not have enough statistical power due to the smaller number of the severe group. Finally, this was an observational study, and although we adjusted for potential confounders as much as possible, we could not exclude residual confounders when evaluating the association between the severity of AH and the outcome.


This multicentre study indicated that elderly patients and indoor occurrence of AH accounted for approximately 80% of our registry. One-quarter of AH cases were associated with death, although no significant difference in mortality was observed based on the severity of the AH. Active prevention and intervention should occur for this important public health issue.


We are deeply indebted to all members of the J-Point registry group for their contribution. We also thank Ms Marie Hara for study support.



  • Contributors TM contributed to the study concept and design, analysis and interpretation of the data, drafting of the manuscript and critical revision of the manuscript. SM, NE, NM, YohO, TJ, YS, NO, MW, MN, AT, YF and YosO collected the data and contributed to the study concept and design and the interpretation of the data, and the critical revision of the manuscript. TK and BO contributed to the study concept and design, and the critical revision of the manuscript. All authors read and approved the final manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval Kyoto Prefectural University of Medicine.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles