We conducted a retrospective cohort study of all consecutive adult (≥15 years) patients admitted to the medical ED at Odense University Hospital over a one-year period (1 September 2010 through 31 August 2011). The medical ED serves a population of 235,000 adults and as a medical admission unit for the following medical specialities: general internal medicine, infectious diseases, gastroenterology, geriatric medicine, rheumatology, endocrinology and respiratory medicine. The medical ED receives all acutely admitted medical patients from either primary care or from the hospitals open mixed medical-surgical ED.

Initial vital signs were routinely collected upon arrival to the hospital (blood pressure, heart rate, respiratory rate, rectal temperature, peripheral oxygen saturation, recorded oxygen therapy, and Glasgow Coma Scale). Any later vital signs were collected as deemed necessary by the attending physician, but all vital signs were routinely recorded at least every eight hours. There were no interventions related to the study, and all patients received care following the standard operating procedures. We excluded patients unidentified throughout the entire course of admission, patients without a Danish civil registration number, and patients without any recorded vital signs or with initial vital signs recorded >2 hours after arrival to the hospital. We included only the first admission of each patient, within the study period.

We defined normal vital signs as systolic blood pressure >100 mmHg, heart rate ≥50/min and ≤90/min, respiratory rate ≥8/min and ≤20/min, rectal temperature ≥36.0° and ≤38.0°Celsius, oxygen saturation ≥94% (≥90% in patients with a history of chronic obstructive pulmonary disease [COPD]) and no need for oxygen therapy at arrival to the hospital and a Glasgow Scale Score of 15 or registered as awake and alert. Clinical deterioration was defined as vital signs beyond the definition of normal.

Immunosuppression was defined as one or more of the following identified before the current admission:

1) A moderate to high intake of immunosuppressant medication (corticosteroids or antineoplastic and immunomodulating agents) 2) Primary immunodeficiency diseases; 3) Acquired Immune Deficiency Syndrome defining diseases; 4) A newly diagnosed malignancy registered in the Danish Cancer Registry, up to a year prior to admission; 5) Diagnosis of organ transplantation.

Alcoholism-related condition was defined as at least one of the following before the current admission:

1) A redeemed prescription of Disulfiram from 2007 and onwards; 2) ≥2 admissions with a discharge diagnosis of an acute alcohol episode; 3) At least one admission with a discharge diagnosis of a chronic alcohol related diagnosis; 4) A registration in The Danish National Register of Alcohol Abuse Treatment.

Comorbidity was reported by Charlson Comorbidity Index based on discharge diagnoses in the last 10 years and up to seven days prior the current admission.

Patients were coded as Do-Not-Attempt-Resuscitation (DNAR) if they had a DNAR order placed in the medical record within 24 hours of presentation. If a patient was characterised as having a terminal illness or moribund, we defined them as having a DNAR order as well.

Vital signs registered upon arrival to the department and in the following 24 hours were extracted electronically from the patient records. For each patient, we extracted data from the Danish National Patient Register [11] (comorbidity and discharge diagnoses), the Danish Civil Persons Register [12] (vital status [dead or alive] and emigration status), Odense University Pharmacoepidemiological Database [13], the Danish National Cancer Register [14] and the Danish National Alcohol- and Drug Treatment Register, the last three to identify immunosuppression and prior alcoholism-related conditions. All patients were followed up until death, emigration from the country or 30 days after admission, whichever came first.

In compliance with Danish law, the study was notified to and approved by the Danish Data Protection Agency (J No 2008-58-0035), and the access to patient clinical records was approved by the Danish National Board of Health (J No 3-3013-35). No further ethical approval, or consent from participants, is needed for register-based studies in Denmark. Data were anonymised and de-identified prior data analysis.

The reporting of this study conforms to the STROBE statement [15].

Data were presented as medians with 5th and 95th percentiles or proportions with 95% confidence intervals (CI), assuming a binomial distribution. Univariate analysis of baseline characteristics was performed using Chi-squared tests or Mann–Whitney U-tests. A multivariable logistic regression analysis was computed using deterioration as the dependent variable and several predefined potential risk factors for developing clinical deterioration within the first 24 hours after arrival to the hospital, as independent variables. We included each independent variable in the model without assessing their individual association with future clinical deterioration. In each patient we included the first recording of a deterioration among the individual vital signs (systolic blood pressure, pulse rate, GCS, respiratory rate, oxygen saturation and temperature). We computed Kaplan-Meier survival plots of the individual vital signs, where the event was the time of clinical deterioration. Thirty-day mortality was presented as proportions as well as a Kaplan-Meier failure plot and log rank test in a univariate analysis. If a vital sign was missing in the reassessment of the patient, it was defined as if it was within normal range. Statistical analyses were performed with Stata version 13.0 (Stata Corporation LP, Texas, USA).

A total of 6257 patients were admitted to the medical ED within the study period.,4292/6257 (68.6%) had a full set of vital signs at first presentation,1440/4292 (33.6%) had all normal vital signs and were included (Figure 1). Median age among the included patients was 64 years (21–90 years) and 44.0% were male. The distribution of included patients with comorbidity in the Charlson Comorbidity Index categories 0, 1–2 and >2 were 706 (49.0%), 311 (21.6%), 423 (29.4%), respectively. 446/1440 (31.0%) patients had deteriorating vital signs within the first 24 hours after arrival to the hospital.

Independent risk factors for clinical deterioration within the first 24 hours after arrival to the hospital included patients in the age category 65–84 years (odds ratio [OR] 1.79 [95% CI: 1.27–2.52]); patients in the age category 85+ years (OR 1.67 [95% CI: 1.10–2.55]); patients admitted from the open general ED (OR 1.35 [95% CI: 1.07–1.71]) and patients with a DNAR recorded within the first 24 hours after arrival to the hospital (OR 3.76 [95% CI: 1.37–10.31]) (Table 1).

Patients with a discharge diagnosis from the respiratory system had the highest risk of deterioration, (98/153, 64.1%, 95% CI: 55.9–71.6%) followed by discharge diagnoses from the cardiac/circulatory system 30 (35.3% [25.2%–46.4%]) (Table 2).

The overall 30-day mortality of the included patients was 3.8% (95% CI: 2.8–4.9%). In the non-deteriorating group, 19/994 (1.9%, 95% CI: 1.2–3.0%) patients died compared to 35/446 (7.9%, 95% CI: 5.5–10.7%) patients in the deteriorating group (Figure 2), unadjusted hazard ratio 4.11 (95% CI: 2.38–7.10). Of all deteriorating patients, 18/314 (5.7%, 95% CI: 3.4–8.9%) experiencing a single vital sign deteriorating within 24 hours after arrival died within 30 days after admission, 12/99 (12.1%, 95% CI: 6.4–20.2%) with two individual vital sign deteriorating and 5/33 (15.2%, 95% CI: 5.1–31.9%) with three or more individual vital sign deteriorating.

The six individual vital signs of interest contributed with 533 deteriorations among the 446 patients experiencing clinical deterioration. The most common vital sign to deteriorate was the pulse rate (194/533 deteriorations, 36.4%) followed by oxygen saturation (187/533 deteriorations, 35.1%) (Figure 3).

Oxygen saturation expressed the shortest median time to registered deterioration of 7.5 hours, (2.4–19.0 hours) whereas temperature expressed the longest median time of 12.5 hours (3.3–21.1 hours). For a detailed overview of the time of each individual vital sign deterioration, see Figure 3.

The main strength of our study is the large consecutive cohort of acute medical patients with complete follow-up for 30-day all-cause mortality. The main limitation is the absence of standardised time intervals for re-evaluation of the patients. This introduces a bias as the nurses might register vital signs more frequently in some patients (e.g. presenting at increased risk). The current work was a single-centre study from a medical ED at a university hospital, which potentially could affect the generalisability of the study. However, the hospital is the only hospital in the area and is a primary hospital for all residents in the hospital catchment area, in parallel to highly specialised tertiary functions.

The study was conducted at a medical ED describing acutely hospitalised medical patients so the risk factors’ strength of association might be different among patients hospitalised to an abdominal surgery department, cardiology patients or patients in active chemotherapy.

There is always the possibility of residual confounding, which could lead to different estimates when interpreting the independent risk factors in the multivariate models.

Due to the small number of patients died within 30 days after admission we were not able to perform subgroup analysis of mortality in the individual vital signs as well as in the different discharge categories.