Quality of life effects of antithrombin III in sepsis survivors: results from the KyberSept trial [ISRCTN22931023]

Introduction Treatment of sepsis is aimed at increasing both the duration and quality of survival. A long-term focus on quality of life (QoL) in clinical trial evaluations of sepsis care should be a priority. Method QoL data were used to evaluate the effects of intravenous antithrombin III treatment for severe sepsis measured for up to 90 days during the follow-up phase of the KyberSept phase III clinical trial. A visual analog scale and a Karnofsky scale were used to measure physical, psychologic, and social QoL at regular intervals. Changes from baseline between placebo and antithrombin III groups were assessed using Wilcoxon statistical tests, with additional analyses by severity of illness and admitting diagnosis. Results Among all sepsis survivors in the trial, there was a significant advantage on some attributes of QoL in the antithrombin III subgroup of patients who did not receive heparin as compared with the corresponding placebo group. Discussion The present study represents the first attempt to evaluate patient QoL over a relatively long period in a large, randomized, placebo-controlled sepsis trial. Over a 90-day period, survivors of severe sepsis receiving antithrombin III experienced significant improvements as compared with placebo on several attributes of QoL. In conclusion, the present study demonstrated that clinical improvements over an extended time period with antithrombin III were complemented by improvements in QoL, particularly in social and psychologic functioning, in many patients.


Introduction
In treating patients with severe sepsis, physicians are mainly concerned with biologic, physiologic, and clinical outcomes. Of those patients who survive to hospital discharge, little is known regarding their physical functioning (ability to conduct activities of daily life such as walking around), psychologic functioning (mental well-being), and social functioning (communication and relationships with others) after sepsis.
QoL in clinical trial evaluations was among the recommendations of a recent sepsis international advisory group [1].
This report presents the results of an evaluation of QoL among sepsis survivors, done as part of the phase III multicenter KyberSept trial. To the best of our knowledge, limited information on the effects of sepsis treatment on QoL has been reported [2,3]. The report focuses on the effects of antithrombin (AT) III therapy related to physical, mental, and social functioning of patients who were randomly assigned to either high-dose intravenous treatment with Kybernin P ® (Aventis Behring, Marburg, Germany; a plasma-derived AT III concentrate, 30,000 IU over 4 days) or placebo during up to 90 days of follow up. In the trial, overall survival tended to favor AT III over placebo after 90 days of follow up, although the differences were nominal. However, the AT III group that did not receive concomitant heparin exhibited a nominally statistically significant survival advantage over placebo at 90 days of follow-up. The purpose of this investigation was to establish whether these trends also held for the subjective effects of AT III in severe sepsis. It concludes that AT III treatment is effective in improving long-term patient QoL, and that future outcome research studies of severe sepsis should follow up patients throughout their hospital stay and after discharge.

Quality of life instruments
There are a number of important conceptual and methodologic issues in assessing QoL in people who are critically ill, not least of which is the question of how it should be defined. Recent attempts to define QoL have resulted in the development of a functional definition that is measurable, evaluable over time, and readily applied to patients over a wide range of illness severity. Most attempts incorporate the domains of physical, psychologic/cognitive, and social functioning. Each of these domains can be measured in two dimensions: objective assessments of functioning or health status; and more subjective perceptions of general health. The patient's subjective experience translates that objective assessment into the actual QoL experienced. Thus, where the term 'QoL' is used in the present study, it refers to a composite of objective functional impairment and subjective perceptions and expectations. The instruments chosen for the study were designed to reflect this.
QoL gives information on what medical care has achieved for the patient, but severe sepsis is a difficult area in which to obtain such information. Normally, QoL data should be obtained directly from the patient. Unfortunately, patients hospitalized with severe sepsis may be unable to complete a QoL measure or it may be too burdensome. Rather than lose information on patients, the physician investigator was used as a proxy respondent.
Two instruments were used to cover the objective and subjective dimensions. The objective component in the study was measured using the Karnofsky performance scale. The Karnofsky scale emphasizes physical performance and dependency; it is a descriptive, ordinal scale that ranges from 100 (good health) to 0 (dead). Trial investigators assigned percentages based on physical performance ( Table 1). The Karnofsky scale is designed to assess independent functioning and appears to have substantial validity as an indicator of overall physical status. The validity and reliability of the scale have been shown in several populations [4][5][6].
The subjective component of the trial was measured with multiple items using a visual analog scale [7]. A visual analog scale is a line with defined end-points on which investigators indicate a patient's health state. Thus, the investigator placed a mark along a 100 mm line that best described his or her assessment of the patient with the domain in question. There were six domains ( Table 2). The scales ranged from 0 (worst health status) to 100 (best health status). The visual analog scale has been used in several studies of QoL [8][9][10]. Reliability estimates for visual analog scaling items range from 0.40 to 0.95 [11], and these estimates compare favorably with those for other scales [12,13].
The six domains of the first visual analog scale and the Karnofsky score are hereafter referred to as 'attributes'.

The clinical trial
The KyberSept trial was a large, international, phase III clinical trial that enrolled 2314 patients who were evaluable for efficacy and safety in a randomized, double-blind, placebo-controlled design in order to determine the role of high-dose AT III in patients with severe sepsis. Patients randomly assigned to the AT III group received a total of 30,000 IU plasma-derived AT III (Kybernin P ® ) administered as a loading dose of 6000 IU given over 30 min, followed by a continuous infusion of 6000 IU/day for 4 days. The study protocol permitted investigators to prescribe unfractionated or low-molecular-weight heparin for venous thrombosis prophylaxis (≤10,000 IU/day subcutaneous) and heparin flushes for vascular catheter potency (≤ 2 IU/kg body weight per hour intravenous).
In the trial, the all-cause mortality rate at 28 days was almost identical between the placebo group and the group that received AT III (38.7% versus 39.9%; not significant). At the 90-day time point, the AT III group had a nominally lower mortality than did the placebo group (46.4% versus 48.5%). Among the 680 patients in the trial who did not receive heparin concomitantly, there was a statistically significant difference between the AT III and placebo groups. In this subgroup, mortality at 90 days in the AT III subgroup was 44.9%, versus 52.5% in the placebo subgroup (P nominal = 0.03). A complete description of that study and results are reported elsewhere [14].

Outcomes
The Karnofsky scale and visual analog scales were both administered by physician investigators when patients enrolled in the trial (referred to as 'baseline'); during the trial at 28 days, 56 days and 90 days after enrollment; and at discharge from the hospital (if it occurred other than at one of those three intervals). For example, if a patient survived and was discharged at day 60, there would be four assessments (i.e. baseline and days 28, 56, and 60). The primary outcome was change in attribute scores until 90 days after patients were assigned to either treatment or placebo. For this analysis, two patient populations were used. The smaller population was the group remaining in the hospital 90 days from baseline -the 'in-hospital survivors'. The larger population comprised all survivors at 90 days, or at the time point closest to 90 days. This group is labeled the 'all-survivors' group. It includes the in-hospital group. In cases in which there was no patient response at 90 days because they had been discharged, the data from the last assessment but before 90 days were used. This method is equivalent to using the last observation carried forward.
The last observation carried forward method uses the last observed value for that case, and it therefore assumes that the outcome remains constant at the last observed value after discharge. Otherwise stated, it is assumed that QoL in sepsis patients is stable on average for the short time period between discharge and 90 days. It is not possible to conclude firmly that the change in QoL between discharge and 90 days was small and unbiased. However, the fact that the hospital discharge curve between AT III and placebo was almost identical provides evidence that the comparison between treatment groups should be unbiased. Moreover, the physician investigator, the patient, and the family were blinded to the treatment assignment (placebo versus AT III) throughout the duration of the clinical trial.
The secondary outcome was change in the attribute scores at 28, 56, and 90 days after assignment to treatment or placebo. That outcome measure was used in order to compare relative changes between AT III and placebo groups at the 28-day, 56-day, and 90-day assessments.

Patient characteristics
Mean, standard deviation, and range values (for age) and percentages for relevant patient characteristics (at enrollment and administration of heparin) are reported. The overall trial population is presented, as well as the all-survivors and the in-hospital survivors groups. The other variables used were Simplified Acute Physiology Score (SAPS) II [15] (a measure of severity of illness), admitting diagnosis, and concomitant use of heparin. In the SAPS II system, a score-to-risk transformation developed by Le Gall et al. [16] for sepsis patients was used. The risk intervals were identical to the SAPS II strata defined in the trial.

Changes in quality of life between baseline and 90 days
For both the in-hospital survivors and the all-survivors groups, mean changes and standard deviations between baseline and 90 days are shown. Differences in QoL between AT III and placebo were estimated for each attribute. Two-sample Wilcoxon statistical tests were used to determine whether the changes were different between treatment groups.

Differences among patient subgroups
Similarly, Wilcoxon tests were used to identify whether changes in the QoL scores differed for subgroups between Available online http://ccforum.com/content/6/4/349 baseline and 90 days. This analysis was done on the all-survivors population.

Changes in quality of life between baseline and 28, 56, and 90 days
For the all-survivors and in-hospital survivors subgroups, comparative differences in changes between baseline and 28, 56, and 90 days were estimated. For each attribute, the difference in mean change between treatment groups is presented, together with two-sided 95% confidence intervals, so that the differences between placebo and AT III over time are clearly shown.

Statistical considerations
Nominal P values should be regarded as descriptive because no formal null hypothesis was prespecified and no type I error probability can be indicated. No adjustment of P values for multiple testing or multiple confidence intervals was made. P values less than or equal to 0.05 were considered statistically significant. P values above 0.05 but less than or equal to 0.10 were considered indicative of a trend. Confidence intervals for the difference between mean changes were constructed assuming a normal distribution of changes from baseline. Table 3 shows baseline patient demographic data and heparin administration for the 2314 patients who entered the KyberSept trial. The all-survivors group using the last observation carried forward analysis totaled 897 patients, whereas the in-hospital group totaled 118. The patients included in this study population were from a broad mixture of countries.

Patient characteristics
The mean ± SD age of the overall population was 58 ± 17 years in the placebo group and 57 ± 17 years in the AT III group. Mean age was lower (53 ± 17 years) in the allsurvivors population. A clear majority of patients were men (61.5% men versus 38.5% women). In the three populations (overall, all 90 day survivors, and 90 day in-hospital survivors), the sex distribution for the AT III groups was similar. There was a shift toward the lower risk strata of SAPS II score from the overall population to the all-survivors population, whereas the risks in the in-hospital population were slightly higher than  those in the overall population. In the all-survivors group respiratory disorders were the most common diagnostic admitting category, whereas digestive causes of sepsis were the most common among the in-hospital group.
Among all patients enrolled in the trial, 70% received heparin concomitantly. Among the all-survivors, placebo and AT III differed by 3% with respect to concomitant heparin administration. In this population 71% of placebo patients were administered heparin concomitantly; 68% of those administered AT III received heparin. Among the in-hospital survivors at 90 days, 87% in the placebo group received heparin as compared with 70% of those receiving AT III. Comparing AT III groups across populations, the use of concomitant heparin was nearly the same.

Mean changes between baseline and 90 days for the attributes of survivors in the KyberSept trial: patients in hospital at 90 days and all-survivors
In-hospital at day 90 All survivors* Attribute Placebo (mean ± SD) AT III (mean ± SD) P † Placebo (mean ± SD) AT III (mean ± SD) P †  tages for the AT III group as compared with the placebo group in three of the attributes; namely, patient communication and speech, level of alertness, and energy level. In the inhospital population, the differences indicated that communication and speech, level of alertness, and the Karnofsky scale were all more improved for the AT III group than for placebo.
In neither population was the change in patient overall QoL judged to be statistically different. In the in-hospital group, a trend toward greater patient energy levels in the AT III group was found (P < 0.10).

Differences across patient subgroups
As shown in Table 5, although there were generally greater increases in attribute scores in the SAPS II risk for AT III in the all-survivors, none of the differences were statistically significant. There were also no significant differences between placebo and AT III according to the diagnostic subgroups on hospital admission, although a few 'trends' favoring AT III (notably the 'energy level' attribute of QoL) were found. In the heparin subgroups, there was an advantage in some attrib-utes for AT III. Compared with placebo, physical activity was improved significantly more in the AT III subgroup than in the placebo subgroup not receiving heparin. There was also a trend toward greater improvement in three other attributes in the no concomitant heparin subgroup, namely patient mobility, communication and speech, and patient energy level.
Excluding the Karnofsky scores that were unchanged, mean nominal changes in the AT III QoL attributes of the no concomitant heparin group ranged from +4 to +9, whereas those receiving concomitant heparin experienced nominal increases in scores ranging from +2 to +5.

Changes in quality of life between baseline and 28, 56, and 90 days
Figs 1 and 2 show the differences in mean changes in scores between baseline and later periods for the seven attributes measured in the AT III group as compared with placebo. Fig. 1 shows results for the all-survivors population, whereas  Difference in changes between treatment groups in quality of life attributes over time: all-survivors.

Figure 2
Difference in changes between treatment groups in quality of life attributes over time: in-hospital survivors.
dence intervals fully occupying the 'AT III better' side indicate statistically significant improvements in the AT III group. In the all-survivors population, comparative improvements tended to increase over time in the AT III group as compared with the placebo group. An exception was the Karnofsky index, for which the difference between treatment groups remained approximately constant over time. For the in-hospital population, the advantage in the AT III group increased over time for the communication/speech and the level of alertness attributes, whereas the difference between treatment groups remained approximately constant for the other attributes. In the all-survivors, the greatest comparative advantages for the AT III group were found in communication/speech, alertness, and energy level.

Discussion
The present study represents the first attempt to evaluate patient QoL over a relatively long period in a large, randomized, placebo-controlled sepsis trial. The results of the study indicate that, over a 90-day period, survivors of severe sepsis receiving AT III experienced substantial and statistically significant improvements as compared with placebo in several attributes of QoL. These results held for both all-survivors and in-hospital survivors populations. The comparative advantage in QoL for AT III was also maintained from 28 to 56, and 90 days. After 90 days, the study also found an advantage with AT III for some attributes in patients who did not receive heparin concomitantly.
The improvements identified in the study should be interpreted in terms of statistical and clinical significance. Clinical significance deals with the applied value of the change in everyday life. Unfortunately, as with most other QoL instruments, there is not yet a clear standard for 'minimal clinically important differences'. However, improvements by approximately 20% in the in-hospital group of AT III sepsis patients after 90 days in communication/speech and level of alertness probably represent meaningful improvements in many patients.
Another limitation of the present study is the fact that investigators were proxy respondents, rather than patients responding themselves. Like many studies of QoL in critical care settings, physicians were assumed to be sufficiently close to patients to provide valid and reliable data. Although this was not substantiated empirically, there is usually moderate agreement between patients and proxies, although lower levels of agreement have been found in psychosocial functioning [17,18].

Conclusion
Combined with previous findings in 90-day mortality for those without concomitant administration of heparin, this multinational study demonstrates that AT III is associated with meaningful improvements in health for many sepsis patients over an extended 3-month follow-up period. In this subgroup, the clinical improvements over a long time period in patients in the AT III group were complemented by improvements in QoL in many patients, particularly in terms of social and psychologic functioning. Although no significant reduction in 90day mortality was found for the overall population, a nominal advantage in mortality for the AT III group combined with a generally improved QoL profile in the survivors of this group suggests a possible long-term benefit from AT III. Future studies of critically ill patients with severe sepsis or other conditions should include long-term follow up of patients throughout hospitalization and after hospital discharge. In sepsis, long-term follow up appears to provide more meaningful outcome data for patients, families, and society than do standard 28-day, all-cause mortality statistics.

Competing interests
DR, HOK, and SK are employees of Aventis Behring, the latter also owning Aventis stock options. SO is supported by a grant from the Genetics Institute, Cambridge, MA. He also is principal investigator in the PAF Acetylhydorolase study carried out by ICOS, Bothell, WA.
• In the present study, clinical improvements over an extended time period with AT III were complemented by improvements in QoL, particularly in social and psychologic functioning, in a predefined subgroup of patients who did not receive heparin • Combined with previous research on 90-day mortality, the study suggests that AT III may confer long-term benefit in sepsis patients • Among the 'all survivors' group of patients, communication and speech, level of alertness, and energy level exhibited the greatest gains at 90 days • In most cases, significant differences in patient QoL scores at 90 days after hospital admission were also found at the 28-day and 56-day assessments • Future studies in patients with severe sepsis should include long-term follow up throughout hospitalization and after hospital discharge