Article Text
PDF
Abstract
Objective In existing risk stratification and resuscitation guidelines for sepsis, a hypotension threshold of systolic blood pressure (SBP) below 90–100 mmHg is typically used. However, for older patients, the clinical relevance of a SBP in a seemingly ‘normal’ range (>100 mmHg) is still poorly understood, as they may need higher SBP for adequate tissue perfusion due to arterial stiffening. We therefore investigated the association between SBP and mortality in older emergency department (ED) patients hospitalised with a suspected infection.
Methods In this observational multicentre study in the Netherlands, we interrogated an existing prospective database of consecutive ED patients hospitalised with a suspected infection between 2011 and 2016. We investigated the association between SBP categories (≤100, 101–120, 121–139, ≥140 mmHg) and in-hospital mortality in patients of 70 years and older. We adjusted for demographics, comorbidity, disease severity and admission to ward/intensive care using multivariable logistic regression.
Results In the 833 included older patients, unadjusted in-hospital mortality increased from 4.7% (n=359) in SBP ≥140 mmHg to 20.8% (n=96) in SBP ≤100 mmHg. SBP categories were linearly associated with case-mix-adjusted in-hospital mortality. The adjusted ORs (95% CI) for ≤100, 101–120 and 121–139 mmHgcompared with the reference of ≥140 mmHg were 3.8 (1.8 to 7.8), 2.8 (1.4 to 5.5) and 1.9 (0.9 to 3.7), respectively.
Conclusion In older ED patients hospitalised with a suspected infection, we found an inverse linear association between SBP and case-mix-adjusted in-hospital mortality. Our data suggest that the commonly used threshold for hypotension is not clinically meaningful for risk stratification of older ED patients with a suspected infection.
- aged
- emergency department
- geriatrics
- infectious diseases
- risk management
Statistics from Altmetric.com
Key messages
What is already known on this subject
In risk stratification and resuscitation guidelines for ED patients with suspected infection, a static hypotension threshold for systolic BP (SBP) below 90–100 mmHg is typically used because below this threshold prognosis decreases dramatically.
In older patients with a suspected infection, the clinical relevance of SBP in a seemingly ‘normal’ range (>100 mmHg) is still poorly understood, as they may need higher SBP for adequate tissue perfusion due to arterial stiffening.
What this study adds
In older ED patients hospitalised with a suspected infection, we found an inverse linear association between SBP and case-mix-adjusted in-hospital mortality.
Our data suggest that the commonly used threshold for hypotension is not clinically meaningful for risk stratification of older ED patients with a suspected infection.
Introduction
In risk stratification and resuscitation guidelines for emergency department (ED) patients with suspected infection, a hypotension threshold for systolic blood pressure (SBP) below 90–100 mmHg is typically used because below this threshold prognosis decreases dramatically.1
However, for older patients with a suspected infection, the clinical relevance of SBP in a seemingly ‘normal’ range (>100 mmHg) is still poorly understood, as they may need higher SBP for adequate tissue perfusion due to arterial stiffening.
In older trauma patients it has already been suggested that SBPs need to be above 110 mmHg.2 This is in line with recent findings in surgical sepsis patients in whom mortality rates increased dramatically below an SBP of 105 mmHg.3 Importantly, this may partially explain the recent observation that risk stratification and screening tools using cut-offs of 100 mmHg are inappropriate for older ED patients who were hospitalised with a suspected infection.4
Thus, a SBP >100 mmHg falsely interpreted as ‘normal’ may result in delayed recognition and/or resuscitation in older patients with a suspected infection, potentially impacting outcomes.5
In this short report, the association between SBP and case-mix-adjusted in-hospital mortality was therefore investigated in ED patients 70 years and olders who were hospitalised with a suspected infection.
Methods
Study design and setting
This is an observational multicentre study on an existing database of prospectively collected data as part of an ongoing quality improvement programme in three Dutch EDs.
In Leiden University Medical Center, a tertiary care centre with an annual census of ~30 000 patients, data were collected from 1 April 2011 to 1 February 2016. In Rijnstate Hospital, an urban care centre with ~30 000 ED visits per year, data were collected from 1 March 2012 to 1 November 2012. Recently, the Albert Schweitzer Hospital (ASZ), an urban care facility with an annual census of ~25 000 also started participation in the quality improvement programme. In the ASZ data were collected from 1 September 2015 to 1 December 2015.
The study design and quality improvement programme have been described in detail previously.5
Selection of participants
Consecutive patients ≥70 years who were hospitalised with a suspected infection after receiving intravenous antibiotics in the ED were included. The age cut-off of ≥70 years was used because this is in line with the age cut-off used in government-initiated interventions in the Netherlands.
Methods and measurements
Demographics, comorbidities, laboratory values, vital signs, treatment administered (including antibiotics, intravenous fluids and oxygen), disposition from the ED and outcomes were collected as described previously.5
SBP was measured non-invasively with the MP52 monitor (Philips, Eindhoven, The Netherlands). The initial SBP was measured within half an hour after ED registration and divided into four categories(≤100, 101–120, 121–139 and ≥140 mmHg)based on the threshold for hypotension in the quick Sequential Organ Failure Assessment (qSOFA) score and the mean SBP in the general Dutch population of ≥70 years.6 7 Vital sign readings were transferred automatically from the monitor to the digital hospital information system with the PDMS software (Hix, Amsterdam, the Netherlands). The data abstractors verified whether this was a realistic or erroneous reading.
The Predisposition, Infection, Response and Organ failure (PIRO) score was used as a measure of disease severity, taking into account demographic, comorbidities and acute physiology parameters.8 Importantly, the PIRO score has been validated for ED patients with a suspected infection and has better predictive performance than the qSOFA in our patient population.4 8
Directly after ED discharge it was assessed whether a patient was admitted to a normal ward or to a medium or intensive care unit (MCU or ICU), because this reflects response to ED treatment and disease severity not captured in the initial PIRO score. The MCU is a monitored ward similar to a coronary care unit.
The primary outcome measure was in-hospital mortality.
Data analysis
An association as opposed to a prediction model was constructed to investigate the association between SBP categories and in-hospital mortality using multivariable logistic regression, adjusting for PIRO score and MCU or ICU admission.9 Other potential confounders were investigated in various sensitivity analyses (online supplementary file 1). The ORs are reported with 95% CIs. We used the rule of thumb that ~10 events/covariate were needed. We could enter ~4 covariates in the models. All data were analysed using the SPSS V.23.0 software.
Supplementary file 1
Results
Table 1 shows the characteristics of the 833 included patients aged ≥70 years, who represented 35.1% of all 2370 consecutive patients included in the database. In-hospital mortality data were missing in 12 included patients.
Patient characteristics
Unadjusted in-hospital mortality increased with every decrease in SBP category (table 2). In older ED patients the adjusted ORs linearly increased with decreasing SBP categories (figure 1), independent of the type of adjustment for confounding (online supplementary file 1).
The adjusted ORs for in-hospital mortality as a function of systolic blood pressure category. The dashed line with closed triangles represents patients ≥70 years. *Indicates that the adjusted OR is statistically different from the reference systolic BP category (>140 mmHg = closed square). The association between systolic BP category and in-hospital mortality was adjusted for Predisposition, Infection, Response and Organ failure score, and admission to a ward or medium care or intensive care unit. See main text for adjusted ORs.
Unadjusted in-hospital mortality as a function of initial systolic BP category in patients older than 70 years of age
The adjusted ORs (95% CI) for ≤100, 101–120 and 121–139 mmHg compared with the reference of ≥140 mmHg were 3.8 (1.8 to 7.8), 2.8 (1.4 to 5.5) and 1.9 (0.9 to 3.7), respectively. As shown in table 3, the adjusted ORs did not change if the type of hospital was included in the model.
Multivariable logistic regression model with the association between initial systolic blood pressure category and adjusted in-hospital mortality
In online supplementary file 2, it is shown that other haemodynamic markers, like the mean arterial pressure and lactate, were also associated with mortality.
Supplementary file 2
Discussion
The present study highlights that case-mix-adjusted in-hospital mortality linearly increases with decreasing SBP categories in older ED patients hospitalised with suspected infection, at SBPs higher than 100 mmHg. This suggests that on average older patients need SBPs much higher than 100 mmHg and that the baseline values of individual patients should always be taken into account.
Our findings are in line with recent observations in surgical sepsis patients suggesting that SBP thresholds should be higher than 90 mmHg,3 and could explain the poor predictive performance of various screenings and risk stratification tools in older ED patients who are hospitalised with a suspected infection.4
Our study could have three implications. First, the absence of a hypotension threshold in older patients suggests that SBP should be incorporated in risk stratification and screening tools as a continuous rather than a dichotomous variable, or that higher thresholds should be used, based on the baseline SBP for every individual patient. Our study suggests that the qSOFA and the national and modified early warning scores should be age-adjusted with regard to SBP.4
Second, our data suggest that any SBP <140 mmHg increases the odds of mortality. Interestingly, 140 is the mean SBP in the general Dutch population aged ≥70 years.7 Even in earlier stages of sepsis, as included in our study, treatment targets may need to be reconsidered with SBPs much higher than 90–100 mmHg.
Finally, most measurements in clinical settings vary with age and are dynamic during the ED stay. Our findings suggest that the risk for adverse outcomes increases at any level of SBP below what is usually considered normal and that these may be different in age-stratified cohorts.
The main limitation of this study was the relatively small sample size, limiting the range of SBPs and age categories that could be investigated with an appropriate number of events per category. Hence, we acknowledge that it is possible that a threshold could be found if a larger range of SBPs could have been investigated in a cohort with more events, and that the association of interest is different in various age categories.
In conclusion, case-mix-adjusted in-hospital mortality increases linearly with decreasing SBP, suggesting that a hypotension threshold does not exist in older ED patients who are hospitalised with a suspected infection, at least in the range of SBPs investigated. Instead, hypotension should be defined on an individual basis for patients ≥70 years of age based on their baseline SBP rather than a specific threshold.
Our findings should be confirmed in larger prospective cohorts enabling a more accurate correction for confounding and investigation of a larger range of SBPs. Future studies on dynamic algorithms based on sliding cut-offs for individual systemic inflammatory response syndrome (SIRS) and organ dysfunction parameters may better inform clinicians in identifying patients at risk for adverse outcomes earlier.10
Acknowledgments
We are grateful to all the nurses, staff members, senior house officers and residents who were involved in patient inclusion.
Footnotes
Contributors BdG devised and designed the study, collected the data, contributed to the analyses and edited the manuscript. MW collected the data, did the analyses and wrote the manuscript. LB, MvL, MW, AA and LEC collected the data and edited the manuscript. DR and ALS edited the manuscript. BdG takes full responsibility for the study as a whole. All authors have read and approved the 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 The study was approved by the medical ethics committee of the LUMC, which waived the need for individual consent because of the purely observational characteristics of the study.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Primary survey