Abstract
Introduction:
Hypoxic-ischaemic brain injury (HIBI), is a common sequalae following out-of-hospital cardiac arrest (OOHCA), it is reported as the cause of death in 68% of patients who survive to ICU admission, while other patients can be left with permanent neurological disability. Prediction of neurological outcome follows a multimodal approach, including use of the biomarker, neurone specific enolase (NSE). There is however no definitive cut-off value for poor neurological outcome, and little research has analysed NSE and long-term outcomes in survivors. We investigated an NSE threshold for poor short-term neurological outcome and the relationship between NSE and poor neurological outcome in survivors.
Methods:
A retrospective study was conducted of all adult OOHCA patients admitted to the Royal County Sussex Hospital ICU between April 2017 and November 2018. NSE levels, Targeted Temperature Management (TTM), cross-sectional imaging, mortality and GCS on ICU discharge were recorded. Assessment of neurological function after a median of 19 months (range 14–32 months) post ICU discharge was undertaken following ICU discharge and related to NSE.
Results:
NSE levels were measured in 59 patients; of these 36 (61%) had a poor neurological outcome due to hypoxic ischaemic brain injury. Youden’s index and ROC analysis established an NSE cut-off value of 64.5 μg/L, with AUC of 0.901, sensitivity of 77.8% and specificity of 100%. Follow-up of 26 survivors after 19 months did not show a significant relationship between NSE after OOHCA and long-term neurological outcome.
Conclusion:
Our results show that NSE >64.5 µg/L has a poor short-term neurological outcome with 100% specificity. Whilst limited by a low sample size, NSE in survivors showed no relationship with neurological outcome post OOHCA in the long term.
Keywords: NSE, prognostication, OOHCA, neurological outcome, resuscitation
Introduction
Out of hospital cardiac arrest (OOHCA) remains a common and significant cause of morbidity and mortality in the Western world, with an annual incidence in the UK of 53 cases per 100,000 people. 1 Despite developments in intensive care medicine, survival after an OOHCA in the UK remains poor, with less than 10% surviving to discharge.2,3
Hypoxic Ischaemic Brain Injury (HIBI) is seen frequently following OOHCA and can manifest as a range of clinical presentations from seizures, coma, myoclonus and neurocognitive dysfunction to brain death. 4 One study found HIBI was the cause of death in 68% of OOHCA patients who survived to intensive care unit (ICU) admission, but subsequently died in hospital, often following withdrawal of life-sustaining therapy (WLST). 2
Current European Resuscitation Council (ERC) 2021 guidelines state that there are no prognostic tests that are 100% specific for poor neurological outcome. 3 The ERC highlight the importance of accurate prognostication in comatose OOHCA patients and suggest using multi-modal investigations to determine whether to withdraw life support, including clinical examination, neurophysiological studies, biochemical markers and imaging studies. 3 Additionally, early prediction of poor outcome could better utilise the use of finite intensive care resources whilst providing relatives a more definitive answer about prognosis. Multiple aspects of OOHCA management can confound clinical examination used to determine neurological status, delaying accurate prognostication 3 ; these include sedatives, neuromuscular blocking drugs and targeted temperature management (TTM). Therefore clinical decisions are usually made at least 72 h post-arrest, after rewarming and without sedation. 3
Neurone specific enolase (NSE) is a glycolytic enzyme released following neuronal cell death that occurs in HIBI. It is a known neurological prognostic blood biomarker used post cardiac arrest. 3 Typically measured day 48–72 h post cardiac arrest, it is advised to be used in conjunction with clinical examination and other investigations. 3 NSE has a half-life of 30 h and values are thought to correlate positively to the extent of HIBI, thereby making it a good predictor of neurological outcome. 5 There are many advantages to using NSE over other prognostication methods, including its independence from sedatives, low cost and easily interpreted quantitative results. 3 As part of the multimodal assessment the ERC guidelines give a NSE cut off value of >60 μg/L at 48–72 h post ROSC, to predict poor neurological outcome. 3
There is vast research on NSE and neurological prognostication. 6 Most research on NSE and neurological outcome, only consider inpatient mortality from HIBI; but there are now recent studies looking at long-term neurological outcome at 6 months. Data has shown good neurological outcome at 6 months with NSE values found between 33 μg/L and 78.9 ng/ml.7,8 Few studies however have looked at NSE cut-off values in the same cohort to establish neurological outcomes after six months. Long-term outcomes are important to establish, to assist with the identification of those who will require ongoing care and to offer statistical information about quality of life after OOHCA.
In this study we measured NSE in OOHCA patients admitted to critical care, to find a cut off value for poor short term neurological outcome. The same cohort was followed up after a median of 19 months to analyse whether NSE had any relationship with long term neurological outcomes.
Methods
A single centre retrospective study was conducted at the Royal Sussex County Hospital (RSCH), Brighton, UK. The RSCH is a 31 bedded tertiary Intensive Care Unit (ICU) and a regional cardiac centre. Institutional ethical approval was granted for this study as a practice review.
Patients
All adult patients admitted to critical care following OOHCA between April 2017 and November 2018 were included in the study; data was collected retrospectively from the critical care electronic database (iMDsoft Metavision, Sunquest ICE, ONEPACs). Exclusion criteria were; pregnancy, previously known irreversible brain damage or neuroendocrine tumours.
Care was given according to the department’s post cardiac arrest guidelines, based on published 2015 guidelines from the European and UK Resuscitation Council. 9 Standard post-OOHCA management following hospital admission includes; computed topography of the brain, urgent coronary angiography and 24 h of targeted temperature management (TTM). Patients were managed for TTM based on department guidelines, these include, having a cardiac arrest presumed cardiac in origin, requiring intubation, having a GCS of less than 7 and lack of extreme cardio-vascular instability. Temperature was guided by individual physician preference in the range of 34°–36° using external or intravenous cooling as required. Prognostication of these patients encompassed multiple clinical assessments, EEGs and somatosensory evoked potentials (SSEP), as per department guidelines. NSE data was unblinded from doctors, but was not used for clinical decision making by the critical care team during this data collection period.
Demographic data included, age, sex, gender and comorbidities. Data on Utstein criteria were not available. The overall data set consisted of information on the patient’s admission through to ICU discharge. Comprising; initial cardiac arrest rhythm, cause of arrest, length of stay, survival to discharge, GCS on discharge, results of cross-sectional brain imaging, anticonvulsant usage, NSE values, NSE sampling time and TTM usage; including data on length of TTM, average temperature and method of cooling.
Blood sampling
NSE samples were collected from the patients under the discretion of the treating consultant typically 48–72 h post arrest. Reasons for no NSE samples taken include; death within the first 48 h of admission, patients with clinically good neurological outcome not requiring ICU admission or clinicians unintentional omission. NSE samples were analysed in the same laboratory using an Electro-chemiluminescent-Immunoassay (ELICA) kit (Roche Diagnostics), and were screened for haemolysis. 10
Outcomes
The primary aim of this study was to evaluate NSE levels and short term neurological outcome in OOHCA patients. Poor short-term neurological outcome was defined as death during admission from neurological causes. Good short-term neurological outcome was defined as survival to hospital discharge. The records of deceased patients were reviewed by an ICU Consultant to ascertain whether death was from neurological causes. A post-arrest neurological cause of death during admission was attributed retrospectively due to the nature of the study, if there was: a history of OOHCA and GCS was 3–4 without sedation for >24 h, or a CT scan showed evidence of hypoxic encephalopathy and GCS <7 and >4 at >24 h post sedation. If a primary neurological cause, such as subarachnoid haemorrhage, for cardiac arrest and coma with reduced GCS was found this was not included in post-arrest neurological death.
Secondary aims include analysis of NSE and prediction of long term neurological outcome. Long term follow up was achieved using telephone contact with each patient between 14 and 32 months following ICU discharge. Verbal consent was obtained to collect anonymised information. A single investigator, who was blinded to NSE values, conducted all patient interviews. Neurological outcome was stratified using in-house standardised questions and the validated cerebral performance category (CPC). Poor long-term neurological outcome was defined as CPC 3 or over, with good long-term outcome defined as CPC 1 or 2 Patients were asked to subjectively assess their QoL on a locally devised numerical scale from 1 to 5, where 1 was I can do everything and 5 was I am significantly impaired and unable to do most of the things I’d like to be able to do.
Statistical analysis
Data handling, statistical analysis and graphics were performed using SPSS (IBM SPSS statistics, version 25) and Microsoft Excel 2018. Mann-Whitney U tests were used to compare data sets. Receiver-operating characteristic curve (ROC) was performed and area under the curve (AUC) determined. Youden’s index was maximised to define a cut-off value with 100% of specificity for NSE to predict poor neurological outcome. Statistical significance was set at a p value of less than 0.05 (two-sided) for all tests.
Results
Patient characteristics
Eighty-nine OOHCA patients were admitted to the RSCH ICU between April 2017 and November 2018 (Figure 1). Twenty-four patients did not have an NSE sample collected. Of the 65 patients who had an NSE level measured, 42 (65%) died during their admission, 36 (84%) of these died from Hypoxic Ischaemic Brain Injury and were therefore classified as having poor short-term neurological outcome, see Appendix for supplementary data. The remaining 6 (7%) patients died from either non-neurological causes or had unknown/unclassified causes of death and did not meet the criteria to be included in the NSE analysis.
Figure 1.
Study flow chart, illustrating the total number of OOHCA patients included (n = 89), the number who died due to neurological causes during admission (n = 36) and the number of patients who had poor (n = 3) and good (n = 15) long term neurological outcome.
Overall, a total of 59 patients were included in the NSE analysis. Of the 23 patients who survived to ICU discharge 4 patients died and 1 was uncontactable and therefore excluded from long term outcome analysis. Three (17%) patients had poor long term neurological outcome and 15 (83%) patients had good long term neurological outcome based on CPC results.
Characteristics of patients who had an NSE sample measured, excluding death from non-neurological causes, are shown in Table 1. A 36 (61%) died during their admission from Hypoxic Ischaemic Brain Injury neurological causes, with 27 (75%) of those being male. Of the patients with good neurological outcome, 18 (78%) were treated with TTM. HIBI on cross sectional imaging was found in 5 (22%) of those with good neurological outcome and 21 (58%) of those with poor neurological outcome.
Table 1.
Baseline characteristics of patients who had an NSE measured.
| Patients with an NSE measured (59) | Patients with poor neurological outcome (36) | Patients with good neurological outcome (23) | p Value | |
|---|---|---|---|---|
| Male gender, n (%) | 42 (71) | 27 (75) | 15 (65) | 0.606 |
| Average age (years) | 63 | 65 | 61 | 0.706 |
| TTM, n (%) | 32 (54) | 14 (39) | 18 (78) | 0.003 |
| HBI on imaging, n (%) | 26 (44) | 21 (58) | 5 (22) | 0.001 |
| Anti-epileptics used, n (%) | 31 (53) | 20 (56) | 11 (48) | 0.964 |
Expressed as absolute numbers and percentages.
NSE cut off value for poor neurological outcome
Median time from ICU admission to NSE sample sent was 72 h (IQR 24–96 h). All patients who had an NSE value measured over 64.5 μg/L fell into the poor neurological outcome, as shown in Figure 2. Median NSE for those with good neurological outcome was 20 μg/L (IQR 16.5–35.5 μg/L) versus 103 μg/L (IQR 69.5–179 μg/L) for poor outcome (p < 0.001). There was a statistically significant difference between the mean NSE for those who had good neurological outcome and those who did not (p < 0.001).
Figure 2.
Box plot showing NSE values in those who survived to discharge compared to those with poor neurological outcome. Results are expressed as medians and IQR (n = 59).
NSE predicted poor neurological outcome with area under the receiver-operating curve (AUC) of 0.901, as shown in Figure 3. By maximising Youden’s index, Figure 3 shows that an NSE cut off value of 64.5 μg/L (sensitivity of 75%, specificity of 100% and 95% CI) is able to reliably predict poor outcome (p < 0.01). No patients with an NSE value above this cut off point had a good neurological outcome.
Figure 3.
Receiver-operating characteristic curves with AUC for prediction of poor neurological outcome based on NSE values.
The relationship between NSE and long-term neurological outcome
In total, 18 patients were analysed in the long-term follow up. Contact was made median time 19 months (range 14–32 months) post ICU discharge. Three patients had a poor long-term neurological outcome. Four patients died during the follow up period, the cause of which we were unable to attain, and therefore weren’t included in analysis. One patient was lost to follow up. There was no significant relationship found between NSE and neurological outcome (good outcome defined as CPC 1 or 2, poor outcome as CPC ⩾ 3) p = 0.0963, as shown in Figure 4.
Figure 4.
Box plot showing the NSE values for those with good (CPC 1 or 2) and poor long term neurological outcome (CPC ⩾ 3), results are expressed as medians and IQR (n = 18).
Discussion
Our study has shown NSE to be a robust predictor of poor short term neurological outcome. In our patient group we found a cut off value of 64.5 μg/L with 100% specificity (95% CI 74%–100%), 75% sensitivity (95% CI 63%–88%) and AUC of 0.901. No patient with an NSE value above 64.5 μg/L survived to discharge. However, NSE was not found to be a reliable indicator of long term neurological outcome in the patients who survived to hospital discharge.
At present, NSE used in prognostication is limited, as data is not unanimous regarding NSE cut-off values. Early prognostication studies analysed NSE before the introduction of targeted temperature management, finding cut-off values of 33 μg/L for poor neurological outcome.11–13 More recent data looking at NSE and short-term neurological outcome have found much higher NSE cut-off values,7,14–17 with many studies showing similar results to ours.14,15,17,18 In our cohort, eight patients (36%) had values between 33 and 64 μg/L and had good neurological outcome. A recent ERC 2020 review that found an NSE cut off value of >60 μg/L at 48–72 h post ROSC, gave a minimal risk of a false positive prediction and has been implemented in the current post resuscitation 2021 guidelines. 3 Our data therefore adds further weight to the ERC recommendations.
Variability in NSE data in different studies may be due to the use of different analysers, extraneuronal sources of NSE such as neuroendocrine tumours or the change in NSE values within the first 72 h post-arrest.19,20 Rundgren et al. 21 found the results of two NSE analysis methods varied by up to 36%. Our study benefits from being single centred, with NSE measured immediately in one lab using ELISA, limiting the influence of assay variability and laboratory processing. The laboratory used in this study also benefits from haemolysis testing, a variable that is known to drastically influence NSE values. 20 Therefore, further research into standardising assays, the effect of haemolysis and which haemolysis index to use is needed to prevent inaccurately high results and to enable a standardised cut off for all hospitals.
The NSE result in those who survived to discharge showed no statistical or clinically meaningful relationship with long-term neurological outcome post-OOHCA over 19 months after hospital discharge. It is for this reason that we did not combine statistics from the long and short-term outcome group, to give an overall value for poor neurological outcome, but this is something we would aim to do in future larger studies. It is however noteworthy, that the patient with the highest NSE of 64.5 μg/L had the worst long-term outcome, and remained in a vegetative state.
Various studies have found an association between NSE and long-term outcome at 6 months, however all with differing definitions of good long term neurological outcome. The majority of studies analysed CPC at 6 months.8,19,20 Our study looked at outcomes at approximately 19 months (range 14–32 months), a long length of follow up with little current data in NSE prognostication. A confounder of this long-term follow up was the increased likelihood of poor outcomes from other factors. In addition to a limited sample size, this may reflect why a high proportion of patients (21%) were found to have poor long term neurological outcomes compared to previous studies.7,8
Limitations of the study include; sample size, variable times NSE samples were measured post arrest, unblinded NSE data and a large range of long-term follow up times. NSE measured post arrest had an IQR of 24–96 h. ERC guidelines have found the highest association between NSE and outcomes on 72–96 post arrest.9,16 It is noteworthy that a significant proportion of NSE data has been amassed by studies measuring NSE between 24, 48 and 72 h post arrest.15,16,19 Overall, the limitation of varying NSE times measured post arrest is an outcome we believe is the reality of managing investigations in an NHS hospital and continued to analyse this data to reflect this. This study answers a clinical question, and includes elements of human error that would be present if applied in clinical practice, nonetheless our data is supported by a very similar cut off to current clinical guidance.
Furthermore NSE data wasn’t blinded from clinicians. Although the patients treatment, including prognostication, was founded on the departments post cardiac arrest guidelines, which puts emphasis on a multimodal approach, we cannot exclude that clinicians were not influenced by these values. It is notable that the ERC guidelines at the time of the study did not have a NSE threshold value, which we felt limited the self-fulfilling prophecy bias risk.
Lastly, there was a large range on long-term follow up times, with a median time of 19 months (range 14–32 months) post ICU discharge. This highlights the difficulty of following up patients over an 18 1months study with various lengths of hospital stay. In future studies we would aim to have a much smaller range of follow up.
Conclusion
We investigated the relationship between NSE levels post OOHCA and short term neurological outcome due to HIBI, and also longer term neurological outcome at approximately 19 months. Our results show that NSE >64.5 µg/L is a good indicator of significant hypoxic ischaemic brain injury with 100% specificity for a poor short-term neurological outcome. The length of follow up is a novel aspect of this study that found no relationship between NSE values and long-term outcomes. We suggest that NSE level after cardiac arrest is a useful addition to making immediate prognostic decisions in these patients, but cannot be used to indicate longer term outcomes and recovery.
Supplemental Material
Supplemental material, sj-docx-1-inc-10.1177_17511437231160089 for The use of neurone specific enolase to prognosticate neurological recovery and long term neurological outcomes in OOHCA patients by Caitlyn Maher, Matthew Cadd, Maya Nunn, Jennifer Worthy, Rebecca Gray and Owen Boyd in Journal of the Intensive Care Society
Acknowledgments
We thank the Royal Sussex County Hospital, Brighton, for providing institutional approval for this study. No funding was required for this study. We also thank the authors, who have made substantial contributions to the following. Caitlyn Maher: Formal analysis, methodology, visualisation and writing – original draft. Matthew Cadd: Formal analysis, visualisation and writing – review and editing. Maya Nunn: Formal analysis, visualisation and writing – reviewing and editing. Jennifer Worthy: conceptualisation, formal analysis and reviewing and editing. Owen Boyd: supervision, project administration, conceptualisation, methodology and reviewing and editing. Rebecca Gray: supervision, project administration, conceptualisation and reviewing and editing.
Footnotes
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article.
Ethical approval: Institutional ethical approval was granted for this study by Brighton and Sussex University Hospital as a practice review.
ORCID iDs: Caitlyn Maher 
https://orcid.org/0000-0003-4754-8228
Matthew Cadd
https://orcid.org/0000-0003-4981-2344
Supplemental material: Supplemental material for this article is available online.
References
- 1. Hawkes C, Booth S, Ji C, et al. Epidemiology and outcomes from out-of-hospital cardiac arrests in England. Resuscitation 2017; 110: 133–140. [DOI] [PubMed] [Google Scholar]
- 2. Laver S, Farrow C, Turner D, et al. Mode of death after admission to an intensive care unit following cardiac arrest. Intensive Care Med 2004; 30: 2126–2128. [DOI] [PubMed] [Google Scholar]
- 3. European Resuscitation Council. European Resuscitation Council and European Society of Intensive Care Medicine guidelines 2021: post-resuscitation care. European Resuscitation Council, https://www.cprguidelines.eu/assets/guidelines/European-Resuscitation-Council-and-European-Societ.pdf (2021, accessed 1 June 2021). [Google Scholar]
- 4. Lemiale V, Dumas F, Mongardon N, et al. Intensive care unit mortality after cardiac arrest: the relative contribution of shock and brain injury in a large cohort. Intensive Care Med 2013; 39: 1972–1980. [DOI] [PubMed] [Google Scholar]
- 5. Haque A, Polcyn R, Matzelle D, et al. New insights into the role of neuron-specific enolase in neuro-inflammation, neurodegeneration, and neuroprotection. Brain Sci 2018; 8: 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Sandroni C, Cariou A, Cavallaro F, et al. Prognostication in comatose survivors of cardiac arrest: an advisory statement from the European Resuscitation Council and the European Society of Intensive Care Medicine. Intensive Care Med 2014; 40: 1816–1831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Kim YJ, Ahn S, Sohn CH, et al. Long-term neurological outcomes in patients after out-of-hospital cardiac arrest. Resuscitation 2016; 101: 1–5. [DOI] [PubMed] [Google Scholar]
- 8. Oksanen T, Tiainen M, Skrifvars MB, et al. Predictive power of serum NSE and OHCA score regarding 6-month neurologic outcome after out-of-hospital ventricular fibrillation and therapeutic hypothermia. Resuscitation 2009; 80: 165–170. [DOI] [PubMed] [Google Scholar]
- 9. Nolan JP, Soar J, Cariou A, et al. European Resuscitation Council and European Society of Intensive Care Medicine 2015 guidelines for post-resuscitation care. Intensive Care Med 2015; 41: 2039–2056. https://www.resuscitationjournal.com/article/S0300-9572(15)00330-5/fulltext#secsect0140 (accessed 1 June 2021). [DOI] [PubMed] [Google Scholar]
- 10. Prieto B, Miguel D, Costa M, et al. New quantitative electrochemiluminescence method (ECLIA) for interleukin-6 (IL-6) measurement. Clin Chem Lab Med 2010; 48: 835–838. [DOI] [PubMed] [Google Scholar]
- 11. Zandbergen EGJ, Hijdra A, Koelman JHTM, et al. Prediction of poor outcome within the first 3 days of postanoxic coma. Neurology 2006; 6: 62–68. [DOI] [PubMed] [Google Scholar]
- 12. Rosén H, Sunnerhagen KS, Herlitz J, et al. Serum levels of the brain-derived proteins S-100 and NSE predict long-term outcome after cardiac arrest. Resuscitation 2001; 49: 183–191. [DOI] [PubMed] [Google Scholar]
- 13. Steffen IG, Hasper D, Ploner CJ, et al. Mild therapeutic hypothermia alters neuron specific enolase as an outcome predictor after resuscitation: 97 prospective hypothermia patients compared to 133 historical non-hypothermia patients. Crit Care 2010; 14: R69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Schrage B, Rübsamen N, Becher PM, et al. Neuron-specific-enolase as a predictor of the neurologic outcome after cardiopulmonary resuscitation in patients on ECMO. Resuscitation 2019; 36: 14–20. [DOI] [PubMed] [Google Scholar]
- 15. Gillick K, Rooney K. Serial NSE measurement identifies non-survivors following out of hospital cardiac arrest. Resuscitation 2018; 128: 24–30. [DOI] [PubMed] [Google Scholar]
- 16. Vondrakova D, Kruger A, Janotka M, et al. Association of neuron-specific enolase values with outcomes in cardiac arrest survivors is dependent on the time of sample collection. Crit Care 2017; 21: 172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Floerchinger B, Philipp A, Camboni D, et al. NSE serum levels in extracorporeal life support patients—relevance for neurological outcome? Resuscitation 2017; 121: 166–171. [DOI] [PubMed] [Google Scholar]
- 18. Storm C, Nee J, Jörres A, Leithner C, et al. Serial measurement of neuron specific enolase improves prognostication in cardiac arrest patients treated with hypothermia: a prospective study. Scand J Trauma Resusc Emerg Med 2012; 20: 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Chung-Esaki HM, Mui G, Mlynash M, et al. The neuron specific enolase (NSE) ratio offers benefits over absolute value thresholds in post-cardiac arrest coma prognosis. J Clin Neurosci 2018; 57: 99–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Mastroianni A, Panella R, Morelli D. Invisible hemolysis in serum samples interferes in NSE measurement. Tumori 2020; 106: 79–81. [DOI] [PubMed] [Google Scholar]
- 21. Rundgren M, Cronberg T, Friberg H, et al. Serum neuron specific enolase: impact of storage and measuring method. BMC Res Notes 2014; 7: 726. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material, sj-docx-1-inc-10.1177_17511437231160089 for The use of neurone specific enolase to prognosticate neurological recovery and long term neurological outcomes in OOHCA patients by Caitlyn Maher, Matthew Cadd, Maya Nunn, Jennifer Worthy, Rebecca Gray and Owen Boyd in Journal of the Intensive Care Society