Acta Anaesthesiol Scand. 2020;00:1–8. wileyonlinelibrary.com/journal/aas
|
11 | INTRODUCTION
Admission to the intensive care unit (ICU) is considered beneficial for patients with or at a high risk of developing serious pathophysiologi-cal derangements or adverse events.1-3 Surgical patients constitute a substantial proportion of ICU patients, yet their outcomes are not well-studied.4 It is not known which variables, including exposure
to surgery prior to ICU admission, are independent risk factors for long-term mortality. This would be a clinically relevant investigation considering the equipoise regarding the value of ICU admission after non-cardiac surgery,5 and has important implications for the alloca-tion, distribution and cost of this limited resource.6
Sweden has the second lowest number of ICU beds per capita in Europe and has a low ICU-to-acute care bed ratio.7 The available Received: 21 November 2019
|
Revised: 24 February 2020|
Accepted: 5 April 2020DOI: 10.1111/aas.13604
O R I G I N A L A R T I C L E
Exposure to surgery is associated with better long-term
outcomes in patients admitted to Swedish intensive care units
Monir Jawad
1,2| Amir Baigi
3| Michelle Chew
4This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2020 The Authors. Acta Anaesthesiologica Scandinavica published by John Wiley & Sons Ltd on behalf of Acta Anaesthesiologica Scandinavica Foundation
1Central Hospital in Kristianstad,
Kristianstad, Sweden
2Lund University, Lund, Sweden 3Gothenburg University, Gothenburg,
Sweden
4Department of Anaesthesia and Intensive
Care, Medical and Health Sciences, Linköping University, Linköping, Sweden Correspondence
Monir Jawad, Consultant Anaesthetist, Central Hospital in Kristianstad, J. A. Hedlund v. 1, 291 33 Kristianstad, Sweden. Email: monirsaeed@hotmail.com Funding information
The study was funded by Region Östergötland, and region Halland County Councils and Linköping University.
Background: Long-term outcomes of patients admitted to intensive care units (ICUs)
after surgery are unknown. We investigated the long-term effects of surgical expo-sure prior to ICU admission.
Methods: Registry-based cohort study. The adjusted effect of surgical exposure for
mortality was examined using Cox regression. Secondary analysis with conditional logistic regression in a case-control subpopulation matched for age, gender, and Simplified Acute Physiology Score III (SAPS3) was also conducted.
Results: 72 242 adult patients (56.9% males, median age 66 years [IQR 50-76]),
admitted to Swedish ICUs in 3-year (2012-2014) were followed for a median of 2026 days (IQR 1745-2293). Cardiovascular diseases (17.5%), respiratory diseases (15.8%), trauma (11.2%), and infections (11.4%) were the leading causes for ICU ad-mission. Mortality at longest follow-up was 49.4%. Age; SAPS3; admissions due to malignancies, respiratory, cardiovascular and renal diseases; and transfer to another ICU were associated with increased mortality. Surgical exposure prior to ICU admis-sion (adjusted hazard ratio [aHR] 0.90; 95% CI 0.87-0.94; P < .001), admisadmis-sions from the operation theatre (aHR 0.94; CI 0.90-0.99; P = .022) or post-anaesthesia care unit (aHR 0.92; CI 0.87-0.97; P = .003) were associated with decreased mortality. Conditional logistic regression confirmed the association between surgical exposure and decreased mortality (adjusted odds ratio 0.82; CI 0.75-0.91; P < .001).
Conclusions: Long-term ICU mortality was associated with known risk factors such as
age and SAPS3. Transfer to other ICUs also appeared to be a risk factor and requires further investigation. Prior surgical exposure was associated with better outcomes, a noteworthy observation given limited ICU admissions after surgery in Sweden.
epidemiological data reveal that Sweden has lower rates of post-operative ICU admission than other countries.8,9 The International Surgical Outcome Study showed that only 1.6% of surgical pa-tients in Sweden were admitted to ICU postoperatively, compared to 9.7% in the rest of the international cohort.9 Swedish postoper-ative outcomes are generally good, arguably because of the provi-sion of some critical care services in post-anaesthesia care unites (PACUs).10
The Swedish Intensive Care Registry (SIR) collects data on pa-tients admitted to ICUs, with 100% coverage of papa-tients admitted to general ICUs since 2015. Published data show that patients ex-posed to surgery prior to ICU admission have better crude ICU mortality rates than other patients.11 This suggests that patients admitted to ICU after surgery may benefit more from intensive care than those not undergoing surgeries. However, a more de-tailed analysis of surgery as an exposure prior to ICU admission has not been made.
This study aimed to determine the long-term outcomes of ICU patients and to investigate its predictive factors. We explored if exposure to surgery prior ICU admission was an independent risk factor for long-term mortality among other predictive variables. Additionally, in a secondary analysis after matching for age, gender and Simplified Acute Physiology Score III (SAPS3) and controlling for confounders, we tested the null hypothesis that there are no differ-ences between long-term survivors and nonsurvivors due to expo-sure to surgery.
2 | MATERIALS AND METHODS
This study was approved by the Regional Ethics Board in Lund, No. 2016/244 in 2016-04-19 and No. 2019-02898 in 2019-05-15. Our request for data retrieval was accepted by the Swedish Intensive Care Registry (SIR) in 2016-05-30. Data on ICU patients are rou-tinely and prospectively uploaded to SIR. Patients and their relatives were informed of the database and its intention and given the option to opt-out of the registry at any time.
We performed a registry-based retrospective observational cohort study using data from SIR between January 1, 2012 and December 31, 2014. Data on all patients aged ≥18 years and ad-mitted to general ICUs with available follow-up status on December 31, 2018 were extracted. The first ICU admission during this period for each patient was defined as the index admission. Direct transfer to another general ICU was regarded as a continuation of the index admission. Only index admissions were included in the analysis. Patients receiving profiled intensive care (eg thoracic and neurosur-gical) were excluded because of their different management path-ways and outcomes.
The primary endpoint was mortality at longest follow-up, cen-sored on December 31, 2018. Only SIR variables with minimal missing data (<10%), clinical plausibility and previous evidence of effect were chosen a priori for analysis. These variables were age, gender, SAPS3, unplanned ICU admission, surgical exposure
(defined as surgery up to 30 days prior to ICU admission), source of ICU admission (ward, accident and emergency department, op-eration theatre, PACU/other ICU, or other sources), main diagnosis group for ICU admission as defined by SIR, transfers to another ICU, and discharge destination of ICU survivors (ward, another hospital, another ICU not included in the database, or home). Data on patient comorbidities or surgical interventions were not avail-able in SIR, and we could not use data on therapeutic interventions or complications during ICU admission due to poor validity and/or completeness. The clinical characteristics and outcomes are pre-sented in numbers and percentages or median and interquartile range (IQR).
A univariate analysis was conducted for each of the predictor variables. The independent predictive power of each covariate was examined using survival analysis with Cox proportional hazards modelling. We build an exploratory model including all variables as covariates with survival from ICU admission to longest follow-up as the outcome. Their individual adjusted effects within the model were calculated and presented as adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs).
To further explore the effect of surgical exposure, we conducted a conditional logistic regression analysis in a case-control subpopu-lation individually matched for age, gender and SAPS3 in a 1:1 ratio. Zero tolerance limits were applied for the matched variables in order to get strict similarity between the groups. Other variables were not included for matching because they were already included in SAPS3 calculation. Cases were defined as patients with a positive primary outcome (dead at longest follow-up date) and controls were cen-sored patients (alive as of December 31, 2018). We also conducted a second, less strict matching procedure, to test the robustness of the first one. Conditional logistic regression was applied in the matched population with ICU transfers, source of ICU admission, surgical ex-posure, unplanned ICU admission, admission diagnosis group, and discharge destination after ICU as predictor variables and mortality at longest follow-up as the outcome. The results are presented as adjusted odds ratio (aOR) and 95% CI.
All tests were double-sided, and the significance level was set at 0.05. IBM® SPSS Statistics® for Windows, Version 24.0. (IBM Corp.) was used for data analysis. All data are reported according to the STROBE guidelines.12
Editorial Comment
Having surgery and then being admitted to the ICU may influence patient risk for important outcomes. In this ret-rospective registry study, analysing more than 72 000 ICU cases from Sweden, for those who had surgery prior to ICU admission, there was a lower mortality risk compared to those who did not have surgery on the way to their ICU admission.
3 | RESULTS
3.1 | Primary analysis
Data from 107 013 admissions were retrieved. After excluding pa-tients with unavailable status on December 31, 2018 (eg papa-tients with no Swedish personal identification number) and those who did not actually receive intensive care (ie those admitted for observation only with no need for organ support), the final cohort consisted of 72 242 patients, Figure 1. The characteristics and outcomes of the whole cohort and matched subpopulation (n = 32 682) are presented
in Table 1. Thirty-day, 1-year and 5-year mortalities in the whole population were 20.2% (14 563/72 242), 31.1% (22 466/72 242) and 44.4% (21 815/49 090), respectively. The follow-up time ranged from 1460 to 2556 days (median 2026 and IQR 1745-2293 days). Nearly one-fifth of the whole cohort was exposed to surgery 30 days prior to ICU admission.
Univariate analyses revealed that all predictor variables ex-cept “unplanned ICU admission” were significantly associated with mortality at longest follow-up [Supplemental Digital Content (SDC) 1]. In Cox regression analysis presented in Table 2, exposure to surgery was independently associated with a decreased hazard
of mortality at the longest follow-up (aHR 0.9; 95% CI 0.87-0.94;
P < .001; Table 2). Further categorization of surgical exposure
ac-cording to the urgency of surgery showed that this effect was seen for both emergency (aHR 0.90; 95% CI 0.86-0.94; P < .001) and
scheduled (aHR 0.91; 95% CI 0.87-0.96; P < .001) surgeries (SDC 2). Admissions from the operation theatre (aHR 0.94; 95% CI 0.90-0.99;
P = .022) or PACU (aHR 0.92; 95% CI 0.87-0.97; P = .003) were also
associated with decreased risk for mortality. Inversely, age, SAPS3,
TA B L E 1 Patient characteristics and outcomes of the whole cohort and the subpopulation of the matched patients
Variables Whole cohort Subpopulation of matched patients
Cohort (n [%]) 72 242 (100) 32 682 (100)
Age (year; median [IQR]) 66 (50-76) 68 (59-75)
Male (n [%]) 41 079 (56.9) 19 186 (58.7)
SAPS3 (point; median [IQR]) 55 (44-66) 55 (48-63)
Transfers to another ICU (once or more) 4181 (5.8) 2081 (6.4)
Source of ICU admission (n [%])
Ward 21 734 (30.1) 10 380 (31.8)
Accident and Emergency Department 37 915 (52.5) 16 329 (50.0)
Operating theatre 7410 (10.3) 3446 (10.5)
PACU or another ICU not included in the database 3588 (5.0) 1803 (5.5)
Others 1595 (2.2) 724 (2.2)
Unplanned ICU admission (n [%]) 68 281 (94.5) 30 599 (93.6)
Surgery within 30 d to ICU admissiona (n [%]) 12 880 (17.8) 6112 (18.7)
Emergency surgery before ICU admissiona (n [%]) 7928 (11.0) 3500 (10.7)
Group of the main diagnosis for ICU admissiona (n [%])
Cardiovascular system diseases 12 660 (17.5) 5743 (17.6)
Respiratory system diseases 11 447 (15.8) 5744 (17.6)
Trauma 8123 (11.2) 2985 (9.1)
Sepsis and infections 8206 (11.4) 3841 (11.8)
Gastrointestinal system diseases 6976 (9.7) 3581 (11.0)
Nervous system diseases 6791 (9.4) 3222 (9.9)
Fluid balance, blood or endocrine diseases 4586 (6.3) 2253 (6.9)
Medical or surgical complications 2219 (3.1) 1069 (3.3)
Renal system diseases 1647 (2.3) 877 (2.7)
Malignancies 981 (1.4) 563 (1.7)
Obstetrics and gynaecological diseases 816 (1.1) 65 (0.2)
Other (eg post-operative, intoxication etc) 7790 (10.8) 2739 (8.4)
Discharge destination of ICU survivors (n [%])
Ward 53 544 (74.1) 25 726 (78.7)
Home 3324 (4.6) 987 (3.0)
Another ICU not included in the database 5275 (7.3) 2638 (8.1)
Another Hospital 3079 (4.3) 1416 (4.3)
ICU length of stay (day; median [IQR]) 1.04 (0.5-2.5) 1.13 (0.6-2.8)
ICU mortality (n [%]) 7020 (9.7) 1915 (5.9)
30-day mortality (n [%]) 14 563 (20.2) 4660 (14.3)
1-year mortality (n [%]) 22 466 (31.1) 8579 (26.3)
5-year mortality (data on 49 090 vs 25 858 patients) (n [%]) 21 815 (44.4) 11 418 (44.2)
Mortality at the longest follow-up (n [%]) 35 668 (49.4) 16 341 (50)
Follow-up (day; median [IQR]) 2026 (1745-2293) 1622 (309-2149)
Abbreviations: IQR, interquartile range; SAPS3, Simplified Acute Physiology Score III; ICU, intensive care unit; PACU, Post-Anaesthesia Care Unit. aas defined by the Swedish Intensive Care Registry
transfer to another ICU, admission diagnosis group (according to SIR definitions) and ICU discharge destination were associated with in-creased mortality.
Unplanned ICU admission was not associated with mortality, but since this was the case in 94.5% of the whole cohort, there was a likelihood of convergence failure in the logistic regression. Therefore, we repeated the Cox regression analysis without “un-planned ICU admission” as a predictor variable. This analysis yielded similar estimates as the original model (SDC 3).
Malignancy and respiratory, cardiovascular, and renal diseases were associated with the highest hazards, while obstetric/gynae-cological diseases were associated with survival. Admissions due to complications were associated with 10% increased hazard of mortal-ity (aHR 1.1; 95% CI 1.02-1.2; P = .02).
3.2 | Secondary analysis
Case-control 1:1 matching for age, gender and SAPS3 yielded a sample of 32 682 patients (16 341 alive and 16 341 deceased patients). Conditional logistic regression analysis in this sub-population showed that surgical exposure was associated with lower mortality (aOR 0.82; 95% CI 0.75-0.91; P < .001; Table 2). In general, the covariates that were associated with increased/ decreased mortality in the unmatched population were also asso-ciated with similar estimates in the matched subpopulation. Since selection bias is an inherent risk when creating matched subpop-ulations, we also tested our hypothesis using another matched subpopulation with less restrictive tolerance limits (±2 years of age and ±2 points for SAPS3 but same gender). This yielded a subpopulation of 36 600 patients. The conditional logistic re-gression analysis of both subpopulations showed almost similar results (SDC 4).
4 | DISCUSSION
Our principle finding was that long-term mortality was influenced by known risk factors such as age and SAPS3, but also modifiable risk factors such as ICU transfers and source of admission. Exposure to surgery up to 30 days prior to ICU admission was associated with de-creased mortality after a median follow-up time of 2026 days. These data provide valuable insights into factors that affect outcomes long after ICU admission in Sweden. Our primary findings were cor-roborated by a conditional regression analysis in a subpopulation matched for three factors that have previously been identified to be important determinants of ICU outcome (age, gender and SAPS3). The finding that surgical exposure prior to ICU admission is associ-ated with better outcomes does not mean that surgery in itself has a protective effect, but is likely to be a reflection of different case-mix between surgical and nonsurgical populations that we were unable to adjust for in our study. However, we show that patients admitted to ICU after surgery have substantial long-term benefits than those
not undergoing surgeries, even after controlling for baseline charac-teristics and illness severity.
Our results are consistent with previous studies that found in-creased long-term survival among surgical ICU patients compared with other ICU subgroups.13 Our analysis confirm the importance of age and SAPS3 as predictive factors for mortality, as shown in pre-vious studies.14-16 We demonstrate that transfer to another general ICU for continuation of intensive care was associated with increased mortality risk, but it did not appear to be related to the number of transfers (SDC 2). Although inter-ICU transfers may represent an epiphenomenon masking patient factors (eg complications needing treatment), this covariate was independently associated with lower survival even when “complications” as an ICU admission cause was included in the multivariable analysis. Inter-ICU transfers have been previously identified as an independent risk factor in a Swedish pop-ulation.17 It would be relevant to further explore the underlying rea-sons for transfer (eg lack of ICU beds or staffing, specialized surgical treatment etc), timing of transfer and number of transfers, in order to further elucidate the risk associated with inter-ICU transfer spe-cifically for surgical patients.
Malignancies and respiratory, cardiovascular, and renal diseases were the most important admission diagnoses associated with mor-tality. ICU admissions due to “medical or surgical complications” were also independently associated with mortality highlighting the risk of failure to rescue in this group of patients.18-20 However, this is inconsistent with the finding that unplanned admissions to ICU were not associated with increased mortality.3 Complications are as-sumed to result in unplanned ICU admissions, and previous studies have identified this to be an important determinant for outcome.21 We speculate that patients with critical care needs (eg increased monitoring and nursing care) prior to manifest organ failures were possibly identified earlier due to protocolized care pathways, good nursing care and critical care outreach teams that are often available in Swedish hospitals.22-24
Of the various sources of ICU admission, patients from moni-tored environments, such as the PACU/ICU or operation theatre, were associated with a decreased mortality risk compared to those admitted directly from the wards. These findings strengthen pre-vious assumptions that being in a monitored environment allows early identification of deranged physiology or organ dysfunction and, thus, timely referral to ICU.25 The clinical relevance of our findings that surgical patients may have better long-term out-comes supports the concept of extending their “intensive care” beyond the borders of those monitored environments for better outcome optimization.26
Since the number of ICU beds is extremely limited in Sweden, only patients with a reasonable likelihood of survival are accepted for intensive care. Although we do not expect patients undergo-ing surgeries to have different admission criteria than those not undergoing surgeries, only about one-fifth of this Swedish ICU population was subjected to surgery prior to ICU admission. This is in contrast with other studies, where almost a half of the ICU population was subjected to surgery prior to admission.27-29 This
difference may be due to the more extensive role of PACU in perioperative patient management in Sweden.10 Swedish PACUs are generally able to provide some invasive monitoring and thera-peutic interventions, which may only be available in ICUs in other countries. These differences between countries should be consid-ered when evaluating our findings.
We chose to conduct a case-control analysis with matching for age, gender, and illness severity, baseline nonmodifiable variables known to be associated with long-term outcomes,30-33 to con-firm the results of Cox analysis.34 Ideally, comorbidity should be
included in the list of matched variables, but we were unable to do this since these data are not collected in SIR. We did not match for admission diagnoses since surgical causes of admissions are inher-ently different from medical causes and matching for this variable would incur a large selection bias. Our criteria for identifying the most appropriate sample were strict, employing zero tolerance levels for each of the matched factors. Although there is always a risk of selection bias in case-control studies, we attempted to mitigate this by nonrepetitive random sampling from the control pool. Estimates of risk from both matched populations confirmed
TA B L E 2 Cox regression analysis of the whole cohort and the conditional logistic regression analysis of the subpopulation of matched
patients
Variables
Whole cohort (n = 72 242) Matched subpopulation (n = 32 682)
aHR 95% CI P aOR 95% CI P
Age (per year) 1.031 (1.031-1.032) <.001
Male 1.016 (0.995-1.038) .144
SAPS3 (per point) 1.028 (1.027-1.029) <.001
Transfers to another ICU 1.331 (1.262-1.404) <.001 1.440 (1.29-1.61) <.001
Source of ICU admission
Accident and Emergency Department 1.016 (0.991-1.041) .205 0.702 (0.59-0.83) <.001
Operating theatre 0.943 (0.897-0.992) .022 0.901 (0.85-0.96) .001
PACU or another ICU not included in the database
0.922 (0.874-0.973) .003 0.869 (0.78-0.97) .015
Others 0.854 (0.790-0.922) <.001 0.914 (0.81-1.03) .153
Ward (reference) 1 1
Unplanned ICU admission 0.963 (0.917-1.013) .142 0.851 (0.76-0.95) .004
Surgery within 30 d to ICU admissiona 0.904 (0.868-0.941) <.001 0.824 (0.75-0.91) <.001 Group of the main diagnosis for ICU admissiona
Cardiovascular system diseases 1.363 (1.287-1.443) <.001 1.003 (0.90-1.12) .956
Respiratory system diseases 1.462 (1.380-1.548) <.001 1.829 (1.64-2.04) <.001
Trauma 0.942 (0.882-1.006) .073 0.669 (0.60-0.75) <.001
Sepsis and infections 1.108 (1.044-1.176) .001 0.954 (0.85-1.07) .431
Gastrointestinal system diseases 1.270 (1.195-1.350) <.001 1.031 (0.92-1.16) .608
Nervous system diseases 1.228 (1.153-1.307) <.001 1.046 (0.93-1.17) .440
Fluid balance, blood or endocrine diseases 1.275 (1.192-1.364) <.001 1.357 (1.20-1.54) <.001
Medical or surgical complications 1.104 (1.016-1.199) .020 0.938 (0.80-1.10) .432
Renal system diseases 1.307 (1.205-1.418) <.001 1.322 (1.12-1.57) .001
Malignancies 2.247 (2.050-2.461) <.001 3.461 (2.77-4.32) <.001
Obstetrics and gynaecological diseases 0.087 (0.041-0.182) <.001 0.086 (0.03-0.24) <.001
Others (reference) 1 1
Discharge destination of ICU survivors
Home 0.793 (0.725-0.867) <.001 0.805 (0.72-0.91) <.001
Another ICU not included in the database 0.874 (0.829-0.921) <.001 0.791 (0.71-0.88) <.001
Another Hospital 0.793 (0.744-0.845) <.001 0.955 (0.83-1.1) .530
Ward (reference) 1 1
Abbreviations: aHR, adjusted hazard risk; CI, confidence interval; aOR, adjusted odds ratio; SAPS3, Simplified Acute Physiology Score III; ICU, intensive care unit; PACU, Post-Anaesthesia Care Unit.
the results of the primary Cox regression, strengthening the con-clusion that patients exposed to surgery have better survival than others.
The most important limitation of this study is that all analyses only extend to the chosen covariates. Since our choice of covariates was limited by SIR data, data regarding comorbidities, Sequential Organ Failure Assessment (SOFA) score, surgical case mix, use of rescue therapies, ICU complications and therapeutic interventions, socioeconomic status, and university hospital status, which are im-portant possible confounders, were not included in the analysis.
Another limitation was defining surgical patients in our cohort. Since it was impossible to audit individual patient records, we used the only available variable defined by SIR as “Surgery undertaken in current hospital admission but less than 30 days before ICU ad-mission.” This is clearly inadequate since some patients may have been admitted to ICU prior to surgery for various reasons such as re-suscitation, hemodynamic optimization, vital organ stabilization, but based on our experience, these reasons constitute only a minority of surgical ICU admissions. Surgical procedures undertaken during or after intensive care are not usually registered in SIR; therefore, little or no data were available on this important clinical variable.
A strength of this study is the almost full coverage of SIR during our inclusion period (95% in 2012, reaching 98% in 2014), com-pleteness and accuracy of the chosen covariates, and validation of these data by SIR.35 All eligible patients were included to minimize selection bias.36 Detection bias was avoided by only including vari-ables with a high degree of accuracy and completeness. Long-term outcomes were available for almost all patients and verified in the Swedish Registry of Deaths.37 Recall bias was not a limitation since data were recorded for other purposes before the occurrence of the outcome and prior to the start of the study.38-40
We stress that our goal was neither to compare surgical vs nonsurgical patients, nor to compare ICU vs no ICU admission of surgical patients. Rather, we aimed only to investigate surgery as an exposure among many other factors that affect outcome. The findings of our study do not imply causality; however, it prompts further investigation into modifiable risk factors, and specific sub-populations that may benefit from intensive care. Therefore, it is immediately relevant as clinicians, health care economists, and strategists struggle to identify patients that will benefit most from intensive care.
Recent studies have not identified any survival benefit from ICU admission after elective surgery.5,22 The current findings neither support nor refute these results, since we did not compare surgical patients admitted vs those not admitted to ICU, a randomization that would likely be ethically unjustifiable. However, we demonstrate that patients who underwent surgery appear to benefit from ICU admission more than those who did not undergo surgery, even after controlling for age, gender, severity of illness and other risk factors that may affect mortality. This difference raises several consider-ations. Firstly, surgical patients seem to take up only a small propor-tion of ICU beds in Sweden despite comparable severity of illness, making them less competitive for such resources. Yet, survival rates
of surgical patients were higher, speaking for the successful disposal of ICU beds in this group of patients.41 Further studies are needed in an expanding surgical population with increasing proportions of elderly and sicker patients.16,42
5 | CONCLUSION
Long-term mortality in Swedish ICU patients is driven by tradi-tional risk factors such as age and SAPS3 score but is also inde-pendently associated with modifiable risk factors such as ICU transfers and source of admission. Exposure to surgery 30 days prior to ICU admission was independently associated with de-creased mortality risk.
ACKNOWLEDGEMENTS
We sincerely acknowledge the contribution and continuous assis-tance we received from the Swedish Intensive Care Registry during the process of completing and cleaning the database.
CONFLIC T OF INTEREST
Monir Jawad and Michelle Chew have received grants to support the conduct of this study from Region Östergötland, Region Halland County Councils and Linköping University. Amir Baigi has no con-flicts of interest.
ORCID
Monir Jawad https://orcid.org/0000-0002-2544-5095
Michelle Chew https://orcid.org/0000-0003-2888-4111
REFERENCES
1. Blanch L, Abillama FF, Amin P, et al. Triage decisions for ICU ad-mission: Report from the Task Force of the World Federation of Societies of Intensive and Critical Care Medicine. J Crit Care. 2016;36:301-305.
2. Farcy DA, Chiu W, Marshall JP, Osborn TM. Critical Care Emergency Medicine. New York, NY: McGraw-Hill Education; 2017.
3. Vlayen A, Verelst S, Bekkering GE, Schrooten W, Hellings J, Claes N. Incidence and preventability of adverse events requir-ing intensive care admission: a systematic review. J Eval Clin Pract. 2012;18:485-497.
4. Rohrig SAH, Lance MD, Faisal MM. Surgical intensive care - current and future challenges? Qatar Med J. 2019;2019:3.
5. Kahan BC, Koulenti D, Arvaniti K, et al. Critical care admission fol-lowing elective surgery was not associated with survival benefit: prospective analysis of data from 27 countries. Intensive Care Med. 2017;43:971-979.
6. Guidet B, Beale R. Should cost considerations be included in medi-cal decisions? Yes. Intensive Care Med. 2015;41:1838-1840. 7. Rhodes A, Ferdinande P, Flaatten H, Guidet B, Metnitz PG, Moreno
RP. The variability of critical care bed numbers in Europe. Intensive Care Med. 2012;38:1647-1653.
8. Pearse RM, Moreno RP, Bauer P, et al. Mortality after surgery in Europe: a 7 day cohort study. The Lancet. 2012;380:1059-1065. 9. International Surgical Outcomes Study group. Global patient
out-comes after elective surgery: prospective cohort study in 27 low-, middle- and high-income countries. Br J Anaesth. 2017;119:553.
10. Jawad M, Baigi A, Oldner A, et al. Swedish surgical outcomes study (SweSOS): an observational study on 30-day and 1-year mortality after surgery. Eur J Anaesthesiol. 2016;33:317-325.
11. The Swedish Intensive Care Registry (SIR). https://www.icure gswe. org
12. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for re-porting observational studies. BMJ. 2007;335:806-808.
13. Brinkman S, de Jonge E, Abu-Hanna A, Arbous MS, de Lange DW, de Keizer NF. Mortality after hospital discharge in ICU patients. Crit Care Med. 2013;41:1229-1236.
14. Juneja D, Singh O, Nasa P, Dang R. Comparison of newer scoring systems with the conventional scoring systems in general intensive care population. Minerva Anestesiol. 2012;78:194-200.
15. Timmers TK, Verhofstad MH, Moons KG, Leenen LP. Long-term survival after surgical intensive care unit admission: fifty percent die within 10 years. Ann Surg. 2011;253:151-157.
16. Revenig LM, Canter DJ, Taylor MD, et al. Too frail for surgery? Initial results of a large multidisciplinary prospective study examining pre-operative variables predictive of poor surgical outcomes. J Am Coll Surg. 2013;217:665-670.e1.
17. Parenmark F, Karlström G, Nolin T, Fredrikson M, Walther SM. Reducing night-time discharge from intensive care. A nationwide improvement project with public display of ICU outcomes. J Crit Care. 2019;49:7-13.
18. International Surgical Outcomes Study group. Global patient out-comes after elective surgery: prospective cohort study in 27 low-, middle- and high-income countries. Br J Anaesth. 2016;117:601-609. 19. Ahmad T, Bouwman RA, Grigoras I, et al. Use of failure-to-rescue to
identify international variation in postoperative care in low-, mid-dle- and high-income countries: a 7-day cohort study of elective surgery. Br J Anaesth. 2017;119:258-266.
20. Massarweh NN, Anaya DA, Kougias P, Bakaeen FG, Awad SS, Berger DH. Variation and impact of multiple complications on fail-ure to rescue after inpatient surgery. Ann Surg. 2017;266:59-65. 21. Gillies MA, Harrison EM, Pearse RM, et al. Intensive care
utiliza-tion and outcomes after high-risk surgery in Scotland: a popula-tion-based cohort study. Br J Anaesth. 2017;118:123-131.
22. Wunsch H, Gershengorn HB, Cooke CR, et al. Use of intensive care services for Medicare beneficiaries undergoing major surgical pro-cedures. Anesthesiology. 2016;124:899-907.
23. Valentini I, Pacilli AMG, Carbonara P, et al. Influence of the ad-mission pattern on the outcome of patients admitted to a respira-tory intensive care unit: does a step-down admission differ from a step-up one? Respir Care. 2013;58:2053-2060.
24. Smith MEB, Chiovaro JC, O'Neil M, et al. Early warning system scores for clinical deterioration in hospitalized patients: a system-atic review. Ann Am Thorac Soc. 2014;11:1454-1465.
25. Tweddell JS, Ghanayem NS, Hoffman GM. All this monitoring… what's necessary, what's not? Semin Thorac Cardiovasc Surg Pediatr Card Surg Annu. 2014;17:81-90.
26. Pelosi P, Ball L, Schultz MJ. How to optimize critical care resources in surgical patients: intensive care without physical borders. Curr Opin Crit Care. 2018;24:581-587.
27. Khuri SF, Henderson WG, DePalma RG, Mosca C, Healey NA, Kumbhani DJ. Determinants of long-term survival after major sur-gery and the adverse effect of postoperative complications. Ann Surg. 2005;242:326-341; discussion 341-3.
28. Rhodes A, Moreno RP, Metnitz B, Hochrieser H, Bauer P, Metnitz P. Epidemiology and outcome following post-surgical admission to critical care. Intensive Care Med. 2011;37:1466-1472.
29. Jhanji S, Thomas B, Ely A, Watson D, Hinds CJ, Pearse RM. Mortality and utilisation of critical care resources amongst high-risk surgical patients in a large NHS trust. Anaesthesia. 2008;63:695-700. 30. Gayat E, Cariou A, Deye N, et al. Determinants of long-term
out-come in ICU survivors: results from the FROG-ICU study. Crit Care. 2018;22:8.
31. Hermans G, van Mechelen H, Clerckx B, et al. Acute outcomes and 1-year mortality of intensive care unit-acquired weakness. A cohort study and propensity-matched analysis. Am J Respir Crit Care Med. 2014;190:410-420.
32. Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: report from a stakeholders' conference. Crit Care Med. 2012;40:502-509. 33. Silva JM, Rocha HMC, Katayama HT, et al. SAPS 3 score as a
pre-dictive factor for postoperative referral to intensive care unit. Ann Intensive Care. 2016;6:42.
34. Brazauskas R, Logan BR. Observational studies: matching or regres-sion? Biol Blood Marrow Transplant. 2016;22:557-563.
35. Sjöberg F, Walther S. Intensive care registries and the evolution of the concept of 'quality of care' - reflections from the 10-year an-niversary symposium of the Swedish Intensive Care Registry. Acta Anaesthesiol Scand. 2012;56:1073-1077.
36. Thygesen LC, Ersbøll AK. When the entire population is the sam-ple: strengths and limitations in register-based epidemiology. Eur J Epidemiol. 2014;29:551-558.
37. Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, Ekbom A. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24:659-667. 38. Docherty AB, Lone NI. Exploiting big data for critical care research.
Curr Opin Crit Care. 2015;21:467-472.
39. Binder H, Blettner M. Big data in medical science–a biostatistical view. Dtsch Arztebl Int. 2015;112:137-142.
40. Prada-Ramallal G, Takkouche B, Figueiras A. Bias in pharmacoepi-demiologic studies using secondary health care databases: a scop-ing review. BMC Med Res Methodol. 2019;19:53.
41. Lindemark F, Haaland ØA, Kvåle R, Flaatten H, Norheim OF, Johansson KA. Costs and expected gain in lifetime health from intensive care versus general ward care of 30,712 individual patients: a distribution-weighted cost-effectiveness analysis. Crit Care. 2017;21:220.
42. Søreide K, Story DA, Walder B. Perioperative medicine and mor-tality after elective and emergency surgery. Eur J Anaesthesiol. 2016;33:314-316.
SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section.
How to cite this article: Jawad M, Baigi A, Chew M. Exposure
to surgery is associated with better long-term outcomes in patients admitted to Swedish intensive care units. Acta
