Supplementary Online Content

Transcription

1 Supplementary Online Content Kaukonen KM, Bailey M, Suzuki S, Pilcher D, Bellomo R. Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, JAMA. doi: /jama emethods. APACHE admission diagnosis and detailed description of statistical analysis efigure 1. Availability of hospital and intensive care unit (ICU) beds efigure 2. All admissions, admissions with severe sepsis and proportion of ICU admissions with severe sepsis efigure 3. Mortality in severe sepsis/septic shock patients and in all other patients efigure 4. Hospital outcomes (death, home, to rehabilitation, to other hospital) in patients with severe sepsis from 2000 to 2012 with numbers indicating denominators efigure 5. Survival according to hospital length of stay efigure 6. Length of stay in deceased patients efigure 7. Stratified odds ratios for mortality in severe sepsis/septic shock referenced against the year 2000 efigure 8. Sensitivity analysis in the 63 ICUs that have contributed data each year from 2000 to 2012 etable 1. Multivariable analysis for risk of being septic etable 2. Characteristics of young patients with severe sepsis etable 3. Mortality in younger patients with severe sepsis/septic shock etable 4.Odds ratios for the annual change in risk in hospital outcomes for all patients with severe sepsis/septic shock in the 63 ICUs that contributed complete data This supplementary material has been provided by the authors to give readers additional information about their work.

2 emethods APACHE admission diagnosis with infection, SIRS criteria and organ failures. APACHE admission diagnosis consistent with infection: nonoperative: pneumonia, parasitic pneumonia, bacterial pneumonia, viral pneumonia, gastrointestinal tract perforation, gastrointestinal tract obstruction, neurologic infection, cellulitis or soft tissue infection; postoperative: respiratory infection, gastrointestinal tract perforation or rupture, cholecystitis or cholangitis, fistula or abscess surgery, peritonitis, cellulitis or soft tissue infection. Systemic inflammatory response syndrome (SIRS) criteria: Body temperature >38 C or <36 C Heart rate >90/minute Respiratory rate >20/minute or PaCO2 lower than 32 mmhg (4.3 kpa) White blood cell count > /µl (>12 x10 9 /L) or <4 000 /µl (<4 x10 9 /L) Organ failures: cardiovascular failure: lowest MAP <65 mmhg or lowest systolic pressure <90 mmhg, respiratory failure: intubation and ventilation, hepatic failure: Bilirubin 5.96 mg/dl (102 µmol/l), renal failure: highest creatinine 3.39 mg/dl (300 µmol/l) or urine output <410 ml/24h or acute renal failure. Acute renal failure was defined as: 1. urine output <410 ml/24h and 2. creatinine 1.50 mg/dl (133 µmol/l) and 3. no chronic dialysis lowest platelet count <50 x10 3 /µl (<50 x10 9 /L)

3 Detailed description of statistical analysis Logistic regression for risk of being septic at ICU admission: Model construction was made using a stepwise selection procedure (inclusion criteria p<0.001) with the following variables considered for inclusion: admission source for hospital and ICU, hospital level, location and care type, year and month of admission, age, gender and chronic comorbidities. The three stage multivariable modelling process to examine the change in hospital outcome over time specifically amongst the sepsis population: Firstly, a stepwise selection procedure was applied to a multinomial logistic model to identify factors that were significantly related to hospital outcome. The following variables were considered for model inclusion: admission source for hospital and ICU, hospital level, size, location, and care type, year and month of admission, age, gender, patient severity, comorbidities, risk of being septic, ventilation status, medical or surgical admission and principal diagnosis. Secondly, first order interaction terms between each significant main effect and year of admission (continuous variable) were created and a stepwise regression procedure was used to sequentially identify interactions that were found to have a statistically significant impact on the relationship between hospital outcome and year of admission. Finally, for increased interpretability, the 15 significant main effects and 5 significant first order interactions with year of admission, were individually fitted to each hospital outcome using ordinary logistic regression. All logistic regression results have been reported as odds ratios (95% CI). Given a large database (>1,000,000 ICU patients, >100,000 sepsis patients), in order to more closely align clinical and statistical significance, a twosided p-value of was used for variable inclusion and to indicate statistical significance. Time to Death Time to death was analysed using Cox-proportional hazards regression accounting for year of admission(treated as continuous), Apache III risk of death, risk of being septic and centre, with results presented as hazard ratios (95% CI). To further clarify the change over time, the 13 year study period was divided into 3 cohorts ( , , ) and presented as a Kaplan Meier curve with a log-rank test for equality across the 3 strata.

4 efigure 1. Availability of hospital and intensive care unit (ICU) beds Available ICU beds/100,000 inhabitants ICU Ventilator beds/100,000 inhabitants Hospital beds/100,000 inhabitants Annual number of intensive care unit (ICU) beds (available beds and ventilator beds, upper panel) and hospital beds (lower panel) per 100,000 inhabitants in Australia and New Zealand.

5 efigure 2. All admissions, admissions with severe sepsis and proportion of ICU admissions with severe sepsis ICU admissions/ 1,000 inhabitants ICU admissions with severe sepsis/ 1,000 inhabitants 12.0 Severe sepsis admissions/all admissions (%) Annual incidence of intensive care unit (ICU) admissions and ICU admissions with severe sepsis per 1,000 inhabitants in Australia and New Zealand (upper panel) and proportion of severe sepsis admissions as percentage of ICU admissions (lower panel).

6 efigure 3. Mortality in severe sepsis/septic shock patients and in all other patients. 40.0% Mortality from 2000 to % 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% All others 13.7% 13.0% 12.5% 12.0% 11.3% 10.8% 10.2% 10.0% 9.7% 9.4% 8.9% 8.7% 8.0% Sepsis 35.0% 33.6% 31.2% 30.1% 28.9% 26.2% 25.7% 24.5% 23.9% 22.0% 21.0% 19.8% 18.4% Crude mortality in severe sepsis/septic shock patients and in all other intensive care unit (ICU) patients.

7 efigure 4. Hospital outcomes (death, home, to rehabilitation, to other hospital) in patients with severe sepsis from 2000 to 2012 with numbers indicating denominators.

8 efigure 5. Survival according to hospital length of stay Hospital survival days Number at risk logrank p<0.001 Kaplan Meier curves for survival in (blue line), (red line) and (green line). Cox porportional hazard ratio for the declining risk in mortality per year (adjusting for APACHE III, risk of being septic and attending site) was 0.97 (95% CI, ).

9 efigure 6. Length of stay in deceased patients Geometric Means (95% Confidence Intervals) for length of stay in deceased patients. Adjustments were performed for APACHE III risk of death and the risk of being septic with patients nested within sites and each site treated as a random effect. There was no significant linear trend for a change over time (P=0.74)

10 efigure 7. Stratified odds ratios for mortality in severe sepsis/septic shock referenced against the year Metro Private Tertiary Rural Odds ratio Small(<300) Med( ) Large(>500) Odds ratio Hos LOS <5 Hos LOS Hos LOS 9 17 Hos LOS >17 Odds ratio Adjustments were performed for APACHE III risk of death and the risk of being septic with patients nested within site and site treated as a random effect. There were statistically significant differences between stratified groups of hospital level (P<0.001) (uppermost panel), hospital size (P<0.001) (middle panel) and hospital length of stay (HOSLOS) (lowest panel). Confidence intervals have been omitted for clarity.

11 efigure 8. Sensitivity analysis in the 63 ICUs that have contributed data each year from 2000 to Odds ratios for hospital outcome referenced against the year P-values for the difference in trajectory for sepsis versus non-sepsis patients with year treated as a continuous variable are as follows; death 0.94 ( ) vs ( ) P=0.43, discharged Home 1.03 ( ) vs ( ) P<0.0001), discharge to other hospital 0.99 ( ) vs ( ) P=0.13 and discharge to rehabilitation 1.08 ( ) vs ( ) P< Odds ratios have been generated adjusting for patient severity and the risk of being septic, with patients nested within hospital and hospital treated as a random effect.

12 etable 1. Multivariable analysis for risk of being septic The variables included in deriving each patient s risk of being septic. Odds ratios (95% confidence intervals, CI) Effect N OR (CI) Effect N OR (CI) ICU Source (ref=operating room) Month (ref=february) ICU Source: Emergency ( ) January ( ) ICU Source: Ward ( ) March ( ) ICU Source: Other ICU same hospital ( ) April ( ) ICU Source: Other hospital ( ) May ( ) ICU Source: Other hospital ICU ( ) June ( ) ICU Source: Unknown ( ) July ( ) Hospital Admission source (ref=home) August ( ) Hospital Admission source: Other Hospital ( ) September ( ) Hospital Admission source: Chronic Care ( ) October ( ) Hospital Admission source: Other ICU ( ) November ( ) Hospital Admission source: Unknown ( ) December ( ) State or Country (ref=victoria) Year of admission (ref=2000) Australia capital territory ( ) ( ) New South Wales ( ) ( ) Northern Territory ( ) ( ) New Zealand ( ) ( ) Queensland ( ) ( ) South Australia ( ) ( ) Tasmania ( ) ( ) Western Australia ( ) ( ) Gender (ref=female) ( ) Male ( ) ( ) Unknown ( ) ( ) Hospital Level (ref=private) ( ) Rural Hospitals ( ) Age group (ref <=44) Metro Hospitals ( ) Age group ( ) Tertiary Hospitals ( ) Age group ( ) Care Type (ref=hdu) Age group >= ( ) ICU ( ) Unknown ( ) Chronic comorbidities (yes vs no) AUROC Cardiovascular disease ( ) Renal disease ( ) Immune suppressive disease ( ) Immune suppressive therapy ( ) Aids ( ) Lymphoma ( ) Metastastes ( ) Leukemia ( ) Immunosuppression ( ) Cirrhosis ( )

13 etable 2. Characteristics of young patients with severe sepsis. Baseline characteristics and outcomes of young ( 44 years) patients with severe sepsis or septic shock stratified by the presence of co-morbidities in APACHE II or III chronic health evaluation co-morbidities. Variable All patients 44 years N=15, without comorbidities b N = 11, with comorbidities b N = 3,678 Age, years, mean (SD) 31.6 (9.6) 30.9 (9.8) 34.0 (8.4) a Male Gender, percent (number) 49% (7524) 48% (5662) 51% (1862) Surgical admission, percent (number) 16% (2450) 18% (2117) 9% (333) a APACHE III Score, mean (SD) 54.8 (29.8) 47.5 (20.7) 59.9 (20.4) a APACHE III risk of death, % median (IQR) 9.1 ( ) 7.3 ( ) 20.5 ( ) a Intensive care unit length of stay, days, median (IQR) 3.0( ) 3.0 ( ) 3.1 ( ) Hospital length of stay, days, median (IQR) 10.8 ( ) 10.0 ( ) 14.4 ( ) a Limitation of treatment, percent (number) 1% (190) 1% (77) 3% (113) a Intensive care unit mortality, percent (number) 9% (1434) 7% (776) 18% (658) a Hospital outcomes Mortality, percent (number) 12% (1784) 8% (902) 24% (882) a Discharge: Home, percent (number) 76% (11724) 79% (9318) 65% (2406) a Discharge: Rehabilitation, percent (number) 3% (514) 3% (411) 3% (103) Discharge: Other Hospital, percent (number) 9% (1449) 10% (1162) 8% (287) a Subgroups Severe Sepsis, percent (number) 54% (8363) 51% (1891) 55% (6472) a Septic Shock, percent (number) 46% (7108) 49% (1787) 45% (5321) a Medical admission, percent (number) 84% (13021) 91% (3345) 82% (9676) a Surgical admission, percent (number) 16% (2450) 9% (333) 18% (2117) a Respiratory failure c, percent (number) 48% (7410) 40% (1479) 50% (5931) a Acute Renal Failure d, percent (number) 10% (1613) 16% (606) 9% (1007) a APACHE II <25, percent (number) 86% (12783) 71% (2488) 91% (10295) a APACHE II 25, percent (number) 14% (2037) 29% (1018) 9% (1019) a APACHE III (<50), percent (number) 56% (6902) 35% (938) 61% (5964) a APACHE III (50-66), percent (number) 22% (2795) 29% (775) 21% (2020) a APACHE III (67-87), percent (number) 16% (1982) 25% (678) 13% (1304) a APACHE III (>87), percent (number) 5% (569) 9% (242) 3% (327) a Sepsis (not UTI e ), percent (number) 23% (3613) 27% (998) 22% (2615) a Sepsis with shock (not UTI e ), percent (number) 19% (3002) 27% (996) 17% (2006) a a Comparisons made between patients without and with co-morbidities (P<0.001), b Co-morbidity as defined by the APACHE II 23 or APACHE III 24 chronic health evaluation classification system c Respiratory failure is defined by mechanical ventilation and intubation. d Acute renal failure is defined by highest creatinine 3.39 mg/dl (300 µmol/l) or urine output <410 ml/24h or 1-3: 1. urine output <410 ml/24h and 2. creatinine 1.50 mg/dl (133 µmol/l) and 3. no chronic dialysis e Urinary Tract Infection

14 etable 3. Mortality in younger patients with severe sepsis/septic shock. Mortality in younger ( 44 years) patients with severe sepsis or septic shock, and subgroups of younger patients. n Mortality 2000 (95% CI) Mortality 2012 (95% CI) Absolute Risk Reduction (95% CI) Relative Risk Reduction (95% CI) All young patients % (18.2%-26.0%) 7.3% (6.1%-8.5%) 14.8% (11.0%-19.1%) 66.9% (58.0%-74.0%) Without co-morbidities a % (12.3%-20.1%) 4.6% (3.4%-5.8%) 11.6% (7.8%-16.0%) 71.6% (60.0%-79.8%) With co-morbidities a % (31.2%-49.6%) 17.2% (13.5%-20.9%) 23.2% (13.5%-33.2%) 57.4% (41.6%-69.0%) Severe Sepsis % (10.4%-19.4%) 5.1% (3.7%-6.5%) 9.9% (5.6%-15.1%) 66.0% (48.5%-77.5%) Septic Shock % (24.2%-37.2%) 9.5% (7.5%-11.5%) 21.2% (14.9%-28.1%) 69.0% (58.5%-76.8%) Medical admissions % (19.4%-28.0%) 8.3% (6.9%-9.7%) 15.4% (11.2%-20.1%) 65.0% (55.2%-72.6%) Surgical admissions % (1.6%-18.0%) 3.0% (1.2%-4.8%) 6.8% (0.8%-18.1%) 69.5% (14.3%-89.1%) Respiratory failure b % (23.1%-33.7%) 11.3% (8.9%-13.7%) 17.1% (11.5%-23.1%) 60.2% (47.6%-69.8%) Acute renal failure c % (35.2%-54.4%) 19.6% (14.9%-24.3%) 25.2% (14.8%-35.6%) 56.3% (40.0%-68.1%) APACHE II < % (8.5%-15.5%) 3.7% (2.7%-4.7%) 8.3% (5.1%-12.4%) 69.2% (54.7%-79.1%) APACHE II % (49.0%-68.6%) 38.7% (31.4%-46.0%) 20.0% (7.6%-31.6%) 34.1% (15.3%-48.7%) APACHE III (<50) % (2.5%-9.9%) 1.1% (0.5%-1.7%) 5.2% (2.2%-10.0%) 83.0% (60.6%-92.7%) APACHE III (50-66) % (2.4%-14.2%) 6.4% (3.9%-8.9%) 1.9% (-3.3%-10.1%) 23.0% (-74.1%-66.0%) APACHE III (67-87) % (13.3%-33.7%) 11.1% (7.0%-15.2%) 12.5% (2.7%-24.3%) 53.0% (16.8%-73.4%) APACHE III (>87) % (50.6%-70.2%) 41.6% (34.2%-49%) 18.8% (6.3%-30.4%) 31.1% (12.5%-45.8%) Sepsis, (notuti d ) % (7.7%-22.9%) 6.0% (3.6%-8.4%) 9.3% (2.6%-18.6%) 60.8% (26.5%-79.1%) Sepsis with shock (not UTI d ) % (29.8%-46.6%) 16.7% (12.4%-21%) 21.4% (12.3%-30.8%) 56.2% (38.7%-68.7%) a co-morbidity as defined by the APACHE II 23 or APACHE III 24 chronic health evaluation classification system b Respiratory failure is defined by mechanical ventilation and intubation. c Acute renal failure is defined by highest creatinine 3.39 mg/dl (300 µmol/l) or urine output <410 ml/24h or 1-3: 1. urine output <410 ml/24h and 2. creatinine 1.50 mg/dl (133 µmol/l) and 3. no chronic dialysis d Urinary Tract Infection

15 etable 4.Odds ratios for the annual change in risk in hospital outcomes for all patients with severe sepsis/septic shock in the 63 ICUs that contributed complete data. Odds ratios (95%CI) for the annual change in risk in hospital outcomes for all patients with severe sepsis/septic shock in the 63 ICUs that contributed complete data from 2000 to 2012 (n=66050 (61%)) and for subgroups of these patients for which the change in risk was found to differ significantly (P<0.001). The annual decline in mortality for all severe sepsis/septic shock in the 63 ICUs that contributed complete data from 2000 to 2012 was given by an odds ratio of 0.94 ( ). This decline in risk was significantly influenced by patient severity [lowest Apache III quartile 0.91 ( ) vs. highest Apache III quartile 0.95 ( )], hospital level [Rural 0.92 ( ) vs. Metropolitan 0.95 ( )], hospital admission source [Home 0.93 ( ) vs. Other ICU 0.97 ( )] and state or country [Western Australia 0.87 ( ) vs. Australian Capital Territory 0.98 ( )]. Effect modifier Outcome: death Outcome: home Outcome: other hospital Outcome: rehabilitation N Overall change per year OR (95%CI) ( ) 1.03 ( ) 0.99 ( ) 1.08 ( ) Patient severity Apache III 1st quartile (<50) ( ) 1.02 ( ) 0.99 ( ) 1.08 ( ) Apache III 2nd quartile (50-66) ( ) 1.02 ( ) 1.00 ( ) 1.06 ( ) Apache III 3rd quartile (67-87) ( ) 1.03 ( ) 1.01 ( ) 1.07 ( ) Apache III 4th quartile (>87) ( ) 1.04 ( ) 1.02 ( ) 1.08 ( ) Intensive Care type ICU ( ) 1.03 ( ) 1.01 ( ) 1.06 ( ) High-dependency unit ( ) 1.01 ( ) 1.01 ( ) 1.00 ( ) Hospital Type Rural ( ) 1.03 ( ) 1.02 ( ) 1.06 ( ) Metropolitan ( ) 1.01 ( ) 1.03 ( ) 1.03 ( ) Tertiary ( ) 1.03 ( ) 1.00 ( ) 1.10 ( ) Private ( ) 1.01 ( ) 0.95 ( ) 1.10 ( ) Hospital Admission source Home ( ) 1.03 ( ) 1.01 ( ) 1.06 ( ) Other Hospital ( ) 1.04 ( ) 0.99 ( ) 1.06 ( ) Chronic Care ( ) 0.98 ( ) 0.99 ( ) 1.10 ( ) Other ICU ( ) 0.99 ( ) 0.97 ( ) 1.24 ( ) Unknown ( ) 0.98 ( ) 1.03 ( ) 1.21 ( ) State or Country Australian Central Territory ( ) 1.00 ( ) 1.02 ( ) 1.00 ( ) New South Wales ( ) 1.04 ( ) 1.00 ( ) 1.10 ( ) Northern Territory ( ) 1.03 ( ) 1.03 ( ) 0.82 ( ) Queensland ( ) 1.02 ( ) 1.00 ( ) 1.13 ( ) South Australia ( ) 1.06 ( ) 1.00 ( ) 0.98 ( ) Tasmania ( ) 1.10 ( ) 0.92 ( ) 1.10 ( ) Victoria ( ) 0.99 ( ) 1.02 ( ) 1.10 ( ) Western Australia ( ) 0.99 ( ) 1.03 ( ) 1.23 ( ) New Zealand ( ) 1.05 ( ) 1.01 ( ) 1.02 ( )