Monitoring of vital signs is a routine procedure in almost all healthcare systems and departments, especially those dealing with critically ill patients. Vital signs objectively indicate the immediate well-being of patients and have long been considered an imperative component of patient assessment and management. Vital signs may also influence decision-making of healthcare workers, for example, at triage in emergency departments (EDs) for critically ill patients 1. So much importance has been placed in monitoring vital signs that wearable devices for continuous remote monitoring of vital signs have been developed in order to improve methods of vital sign monitoring 2. Vital signs are undeniably crucial in the practice of medicine.
Vital signs are physical signs that indicate the physiological status of an individual, such as heart rate, breathing rate, temperature, and blood pressure. These signs may be observed, measured, and monitored to assess the level of physical functioning of an individual. However, any factor that is important in determining the well-being of patients can be considered as a vital sign; for example, pain score has recently been recognized widely as the fifth vital sign. The Glasgow Coma Scale (GCS) and oxygen saturation (SaO2) have also been proposed as vital signs 3,43,4.
What are considered normal or abnormal vital signs? There have been many debates on this issue. However, no consensus has been achieved as of today 5–75–75–7. Normal vital signs are subjective and vary with age, sex, weight, exercise tolerance, environment, medical conditions, and many other factors.
Although vital signs are highly regarded in the field of medicine, not many studies have been conducted to examine how useful vital signs are in predicting clinical outcomes of patients and which vital signs are the most useful and best at predicting outcomes. Clinical scoring systems that incorporate vital signs and other factors have been proposed to assist in risk assessment of patients and for ED triage 8,98,9.
We aimed to evaluate whether vital signs are, in practice, good predictors of clinical outcomes such as cardiac arrest and death for critically ill patients presenting at the ED. We also aimed to observe which vital signs would be the best at predicting clinical outcomes.
We conducted a prospective observational cohort study on 1025 patients between November 2006 and December 2007 presenting at the ED of a tertiary hospital in Singapore. The Institutional Review Board at the Singapore General Hospital granted approval for this research project with a waiver of patient consent.
The Department of Emergency Medicine, Singapore General Hospital, sees between 300 and 500 emergency patients daily. All patients are initially triaged by a nurse, and those with airway, breathing, and circulation problems, or thought to be possibly unstable and needing close monitoring, are routinely put on vital signs monitoring.
We recorded the vital signs of these 1025 patients taken at triage in the ED on arrival. These patients were then followed up on three clinical outcomes: cardiac arrest in 72 h, admission to ICU, and death within 30 days. This data collection was part of an ongoing study of critically ill patients in the ED 10.
The 1025 patients were priority 1 (PACS 1 and 2) patients as determined by the Singapore Patient Acuity Category Scale (PACS) (Table 1). The Singapore PACS is a nationally accepted triage scale that is used by all public EDs in Singapore. Categorization of patients is according to mechanism of injury, physiological status (vital signs) at triage, anatomical site of injury, and chief complaints. These patients were deemed by the triage nurse to be at risk for adverse clinical outcomes.
Exclusion criteria were as follows: age less than 18 years, pregnancy, and those who presented at the ED with cardiac arrest. Only one set of initial vital signs taken at triage in the ED was used in this study and included pulse rate (PR), systolic blood pressure (SBP), diastolic blood pressure (DBP), respiratory rate (RR), SaO2, and the GCS. We excluded PACS 3 and 4 patients as we wanted to focus on critically ill patients presenting at the ED. The event rate (cardiac arrest and death) would likely be very low in PACS 3 and 4 patients.
The above-mentioned vital signs were used as predictor variables and classified into ‘normal’ and ‘abnormal’ using standardized adult parameters 6,116,11. The normal values for each vital sign were arbitrarily regarded as follows for data analysis: PR, 60–100 beats/min; SBP, 90–140 mmHg; DBP, 60–95 mmHg; RR, 12–20/min; SaO2 95% or more; and GCS, 15 (Table 2). Clinical outcomes such as death, hospital admission, ICU admission, and length of hospital stay were subsequently obtained from review of hospital charts.
Data entry was carried out using a spreadsheet application (Excel 2000, Microsoft Corp., Redmond, Washington, USA) and data analysis using SPSS (SPSS Inc., Chicago, Illinois, USA). Descriptive statistics (mean, SD, or median as appropriate) are primarily presented. Receiver operating characteristic analyses were carried out for each parameter for the outcomes of ICU admission, hospital admission, and death. Vital signs were used as predictor variables in univariate and multivariate analyses for the three clinical outcomes of interest. A multivariate analysis was also carried out using the vital signs. Age was added in the multivariate analysis of data to improve the predictive value of the model. Colinearity between the variables of vital signs was assessed by examining the SEs for inflation. Statistical significance was set at P<0.05.
The mean age of the study population was 62 years, and there was an approximate 1 : 1 ratio of women and men in this study. The proportion of patients of different ethnicities in our study reflects the current ethnic distribution in Singapore well. Most of the cases were nontrauma cases and a large percentage of them were PACS 1 patients. Table 3 presents the characteristics of the study population.
Results from the univariate analysis (Table 2) showed that five of the six vital signs, namely, PR [>100/min; odds ratio (OR) 1.93; 95% confidence interval (CI) 1.08–3.44], SBP (>140 mmHg; OR 0.35; 95% CI: 0.18–0.70), RR (<12/min; OR 27.50; 95% CI: 1.68–450.26), RR (>20/min; OR 3.93; 95% CI: 2.19–7.05), SaO2 (<95%; OR 2.28; 95% CI: 1.22–4.26), and GCS (<15; OR 2.96; 95% CI: 1.60–5.49), were significantly associated with cardiac arrest within 72 h. In addition, five of six vital signs, namely, PR (>100/min; OR 1.80; 95% CI: 1.15–2.81), SBP (>140 mmHg; OR 0.60; 95% CI: 0.36–0.98), RR (>20/min; OR 3.56; 95% CI: 2.23–5.68), SaO2 (<95%; OR 2.39; 95% CI: 1.46–3.90), and GCS (<15; OR 5.16; 95% CI: 3.23–8.23), were associated with death within 30 days. Only PR (<60/min; OR 2.25; 95% CI: 1.32–3.81) and GCS (<15; OR 2.41; 95% CI: 1.63–3.55) were associated with ICU admission. RR, SaO2, and GCS were consistent in predicting both cardiac arrest within 72 h and death within 30 days. Among the three, GCS was also found to be consistent in predicting ICU admission (Table 2). Surprisingly, DBP was found to have no significant association in predicting any of the three clinical outcomes of our study.
In the multivariate analysis component of our study, vital signs were analyzed together with age (Table 4). The results revealed that age had a significant effect on cardiac arrest, death, and ICU admission, although effect sizes were small. PR (>100; OR 1.65; 95% CI: 1.00–2.71), SBP (>140; OR 0.41; 95% CI: 0.21–0.79), RR (>20; OR 2.90; 95% CI: 1.67–5.03), and GCS (<15; OR 5.71; 95% CI: 3.40–9.57) were significantly associated with death within 30 days. Age, together with three other vital signs, namely, SBP, RR, and GCS, were found to be independent predictors of cardiac arrest within 72 h. For ICU admission, age and GCS were found to be independent predictors, with inverse association.
The vital signs together with age have very low sensitivity and high specificity in predicting the three clinical outcomes of interest (Table 4). The sensitivity for predicting cardiac arrest was 11.54%, that for death was 22.73%, and for ICU admission was 12.50%. Similar results were observed for death and ICU admission. The specificity for predicting cardiac arrest was 99.28%, that for death was 97.22%, and for ICU admission was 93.80%.
Our study showed that not all vital signs that we have traditionally regarded as ‘vital’ predict clinical outcomes of patients. Vital signs had high specificity but low sensitivity for clinical outcomes. Our study shows that vital signs provide limited information about the status of individual patients and may result in impressions that are inconsistent with more global assessment of the patient. Clinicians should always remember to treat patients, not numbers. Patients with apparently ‘normal’ vital signs may be more ill than they appear.
Although PR and blood pressure have been shown to be indicators of cardiovascular function 12,1312,13, blood pressure was not a reliable predictor of clinical outcomes in our study. This can be explained by the fact that blood pressure is dynamic and can be affected by many environmental or internal factors such as anxiety, noise, stress, and emotions. In addition, studies have demonstrated many factors that can possibly contribute to inaccurate measurements, such as inappropriate cuff size 14, site of measurement 15, position of arm 16, and position of person during measurement 17.
RR was found to be a strong predictor of clinical outcomes in our study. This finding is similar to many other studies conducted to evaluate RR and its usefulness 18–2118–2118–2118–21. However, several other studies have highlighted limitations of usage of RR associated with inaccurate measurements 22–2422–2422–24.
SaO2 was found to be the strongest predictor of clinical outcomes in our study. This is also similar to many other studies conducted 25,2625,26. Studies have shown that use of pulse oximetry in the ED significantly changed medical treatment and has been recommended for all acutely ill patients presenting at the ED 27. However, there are also a number of studies that have demonstrated that pulse oximetry lacks specificity and is not a specific indicator for serious illness 28,2928,29. Pulse oximetry may also be inaccurate because of factors such as reduced perfusion at site of measurement 30 or reduced blood hemoglobin concentration 31. In addition, healthcare workers must be clear that pulse oximetry is not an indicator of adequate ventilation and does not reflect hypercarbia. Arterial blood gas is still necessary for identifying hypercarbia. Therefore, pulse oximetry will definitely not replace any other vital signs like RR but will merely complement the interpretation of other vital signs.
GCS was found to have modest predictive value for clinical outcomes. Not many studies have been conducted to evaluate GCS in outcome prediction. However, GCS was the only vital sign found to be a significant predictor of all three clinical outcomes evaluated in our study, and only GCS among all the other vital signs was associated with predicting ICU admission. This is a finding that is interesting and requires more evaluation and validation, especially in the ED setting. A possible explanation can be one of the limitations of our study, as only one set of vital signs was used. The subsequent progression of the patients’ condition and vital signs were not taken into account. A previous study showed that, although average vital signs measured over 21 min showed higher discriminatory value than initial vital signs, vital sign temporal trends were still diagnostically weak and did not discriminate between sicker patients and controls 32.
Our study has some limitations. The sample size of 1025 patients is not a large number compared with other studies. Vital signs taken only at triage may not be equally reflective for all patients. We did not track changes in vital signs over time. Recording of vital signs was based on routine clinical practice, and we did not assess inter-rater reliability. Also, some patients arriving by ambulance were given prehospital treatment such as oxygen supplementation or intravenous drip, which can significantly alter their vital signs. In our data, we did not distinguish between those with and without oxygen. The cutoff point of normal and abnormal values for vital signs may not be appropriate for all patients as normal values of vital signs can differ among individuals. We also did not have control over the accuracy of measurement of all vital signs taken by healthcare workers. We assume that all healthcare workers at our hospital are well trained in taking accurate vital signs. We do note that in the ED there are always a number of patients who change triage status during their stay in the ED, and this is a limitation we recognize in our study. Finally, we only observed critically ill patients in this study and assume that our conclusions can be extrapolated to stable patients (e.g. PACS 3, 4).
Further studies need to be conducted to find a better risk assessment method for patients and better prediction of clinical outcomes to allow more effective triaging and improved patient care. We intend to investigate the use of heart rate variability at triage to improve prediction of clinical outcomes 33.
Not all vital signs are useful in the prediction of clinical outcomes. Vital signs had high specificity but very low sensitivity as predictors of clinical outcomes. Clinicians should always remember to treat patients and not numbers. Patients with apparently ‘normal’ vital signs may be more ill than they appear. Other factors need to be considered in the initial evaluation of critically ill patients.
This study was supported by grants from SingHealth Talent Development Fund, Singapore (TDF/CS001/2006), and InfoComm Research Cluster, Nanyang Technological University, Singapore (2006ICT09).
Conflicts of interest
All authors have neither commercial, personal associations nor any sources of support that might pose a conflict of interest in the subject matter or materials discussed in this manuscript.
1. Brown LH, Prasad NH. Effect of vital signs
on advanced life support interventions for prehospital patients. Prehosp Emerg Care. 1997;1:145–148
2. Tay FEH, Guo DG, Xu L, Nyan MN, Yap KL. MEMS Wear-biomonitoring system for remote vital signs
monitoring. J Frankl Inst-Eng Appl Math. 2009;346:531–542
3. Yen PK. Nutrition – a vital sign. Geriat Nurs. 1992;13:52–53
4. Robinson MD, Laurent SL, JM Little. Including smoking status as a new vital sign – it works. J Fam Pract. 1995;40:556–561
5. Koziol-McLain J, Lowenstein SR, Fuller B. Orthostatic vital signs
in emergency department patients. Ann Emerg Med. 1991;20:606–610
6. Van Oostrom JH, Gravenstein C, Gravenstein JS. Acceptable ranges for vital signs
during general anesthesia. J Clin Monit. 1993;9:321–325
7. Tarassenko L, Hann A, Young D. Integrated monitoring and analysis for early warning of patient deterioration. B J Anaesth. 2006;97:64–68
8. Goodacre S, Turner J, Nicholl J. Prediction of mortality among emergency medical admissions. Emerg Med J. 2006;23:372–375
9. Gottschalk SB, Wood D, DeVries S, Wallis LA, Bruijns S. The Cape Triage Score: a new triage system South Africa. Proposal from the Cape Triage Group. Emerg Med J. 2006;23:149–153
10. Liu N, Lin Z, Koh Z, Huang G, Ser W, MEH Ong. Patient outcome prediction with heart rate variability and vital signs
. J Signal Process Syst. 2011;64:265–278
11. Fleming S, Thompson M, Stevens R, Heneghan C, Pluddemann A, Maconochie I, et al. Normal ranges of heart rate and respiratory rate
in children from birth to 18 years of age: a systematic review of observational studies. Lancet. 2011;377:1011–1018
12. Knopp R, Claypool R, Leonardi D. Use of the tilt test in measuring acute blood loss. Ann Emerg Med. 1980;9:72–75
13. Baxt WG, Jones G, Fortlage D. The trauma triage rule: a new, resource-based approach to the prehospital identification of major trauma victims. Ann Emerg Med. 1990;19:1401–1406
14. Nielsen PE, H Janniche. The accuracy of auscultatory measurement of arm blood pressure
in very obese subjects. Acta Medica Scandinavica. 1974;195:403–409
15. Tachovsky BJ. Indirect auscultatory blood pressure
measurement at two sites in the arm. Res Nurs Health. 1985;8:125–129
16. Webster J, Newnham D, Petrie JC, Lovell HG. Influence of arm position on measurement of blood pressure
. Br Med J. 1984;288:1574–1575
17. Fleming SE, Horvath JS, Korda A. Errors in the measurement of blood pressure
. Aust N Z J Obstet Gynaecol. 1983;23:136–138
18. Fieselmann JF, Hendryx MS, Helms CM, Wakefield DS. Respiratory rate
predicts cardiopulmonary arrest for internal medicine inpatients. J Gen Intern Med. 1993;8:354–360
19. Gravelyn TR, JG Weg. Respiratory rate
as an indicator of acute respiratory dysfunction. JAMA. 1980;244:1123–1125
20. McFadden JP, Price RC, Eastwood HD, Briggs RS. Raised respiratory rate
in elderly patients: a valuable physical sign. BMJ Clinical research ed. 1982;284:626–627
21. Subbe CP, Davies RG, Williams E, Rutherford P, Gemmell L. Effect of introducing the Modified Early Warning score on clinical outcomes, cardio-pulmonary arrests and intensive care utilisation in acute medical admissions. Anaesthesia. 2003;58:797–802
22. Hewson PH, Humphries SM, Roberton DM, McNamara JM, Robinson MJ. Markers of serious illness in infants under 6 months old presenting to a children's hospital. Arch Dis Child. 1990;65:750–756
23. Hooker EA, O'Brien DJ, Danzl DF, Barefoot JA, Brown JE. Respiratory rates in emergency department patients. J Emerg Med. 1989;7:129–132
24. Simoes EAF, Roark R, Berman S, Esler LL, Murphy J. Respiratory rate
: measurement of variability over time and accuracy at different counting periods. Arch Dis Child. 1991;66:1199–1203
25. Mower WR, Sachs C, Nicklin EL, Safa P, Baraff LJ. Effect of routine emergency department triage pulse oximetry screening on medical management. Chest. 1995;108:1297–1302
26. Kellerman AL, Cofer CA, Joseph S, Hackman BB. Impact of portable pulse oximetry on arterial blood gas test ordering in an urban emergency department. Ann Emerg Med. 1991;20:130–134
27. Holburn CJ, Allen MJ. Pulse oximetry in the accident and emergency department. Arch Emerg Med. 1989;6:137–142
28. Goldhill DR, McNarry AF. Physiological abnormalities in early warning scores are related to mortality in adult inpatients. Br J Anaesth. 2004;92:882–884
29. Hodgetts TJ, Kenward G, Vlachonikolis IG, Payne S, Castle N. The identification of risk factors for cardiac arrest
and formulation of activation criteria to alert a medical emergency team. Resuscitation. 2002;54:125–131
30. Jubran A. Pulse oximetry. Crit Care. 1999;3:R11–R17
31. Harrison GA, Jacques TC, Kilborn G, McLaws ML. The prevalence of recordings of the signs of critical conditions and emergency responses in hospital wards – the SOCCER study. Resuscitation. 2005;65:149–157
32. Chen L, Reisner AT, Gribok A, Reifman J. Exploration of prehospital vital sign trends for the prediction of trauma outcomes. Prehosp Emerg Care. 2009;13:286–294
33. Ong ME, Padmanabhan P, Chan YH, Lin Z, Overton J, Ward KR, DY Fei. An observational, prospective study exploring the use of heart rate variability as a predictor of clinical outcomes in pre-hospital ambulance patients. Resuscitation. 2008;78:289–297