Article Text

Download PDFPDF
Coupled plasma haemofiltration filtration in severe sepsis: systematic review and meta-analysis
  1. Ian Hazzard1,2,
  2. S Jones3 and
  3. T Quinn4
  1. 1Ministry of Defence Hospital Unit, Friarage Hospital, South Tees Hospitals NHS Trust, Northallerton, UK
  2. 2Faculty of Health and Medical Sciences, School of Health Sciences, University of Surrey, Guildford, UK
  3. 3Department of Population Health, New York University Medical School, New York, NY, US
  4. 4Faculty of Health, Social Care and Education, St George's, University of London & Kingston University, London, UK
  1. Correspondence to Major Ian Hazzard MDHU Northallerton, James Cook University Hospital, Middlesbrough TS4 3BW, UK; ianhazzard{at}


Introduction Coupled plasma filtration and adsorption (CPFA) has been used in the treatment of severe sepsis with the intention of removing the proinflammatory and anti-inflammatory mediators from the systemic circulation. It is believed that this interrupts and moderates the septic cascade, but there is uncertainty about the benefits of this therapy.

Methods A systematic review and meta-analysis were performed to estimate the effects of CPFA on mortality in severe sepsis. The Cochrane CENTRAL Register of Controlled Trials, CINAHL, EMBASE, MEDLINE—EBSCO–Host, MEDLINE and ProQuest, were searched from 1997 to 2013. Randomised controlled trials, prospective cohort studies and retrospective cohort studies were included using the Centre for Reviews and Dissemination (CRD) framework. Data were abstracted using standard pro forma, and studies independently reviewed by two authors to confirm inclusion criteria. Quality of studies and risk of bias were assessed using the Grading of Recommendations, Assessment, Development and Evaluation Working Group (GRADE) and Critical Appraisal Skills (CASP) criteria, respectively. Meta-analysis was performed using Review Manager (RevMan V.5.1) software. The primary outcome was 28-day mortality. Secondary outcomes were mediator adsorption (picograms/mL), mean arterial BP (mm Hg) and oxygenation ratio.

Results 17 studies met the inclusion criteria (n=441 patients, 242 CPFA). 14 studies reported the primary outcome of 28-day mortality. There were 88 deaths in CPFA patients versus 118 in those receiving haemofiltration: OR 0.34 (95% CI 0.24 to 0.13). Point estimates of effect on the secondary outcomes of mean arterial pressure and oxygen ratio favoured CPFA. Studies were small and heterogenous.

Conclusions Evidence for CPFA in severe sepsis is sparse, of poor quality and further research is required, however, this meta-analysis noted improvements in survival rates of those patients treated with CPFA.

  • Sepsis
  • Coupled Plasma Filtration Adsorption

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

  • The review identified that CPFA has the potential to reduce mortality for those patients with a delayed diagnosis or who succumb to immunoparalysis.

  • Direct plasma filtration displayed an effect on MAP, oxygen ratio and proof of the adsorption of cytokines.

  • Highly contaminated wounds are a potential vector for sepsis. Should the military casualty evacuation chain be interrupted, lives could be saved by the implementation of plasma filtration, in military hospitals, to treat sepsis.


Fulminant sepsis is the primary cause of death in the intensive care unit (ICU) in the UK,1 and the cause of 36 800 deaths in 2009 across the UK.2 Mortality rates from sepsis are high, ranging from 60% to 90% at 28 days.2 Out of the 100 000 patients admitted into a UK ICU each year, 30% are being treated for sepsis at an estimated cost of £2.5 billion annually.2 ,3

The principal treatment of sepsis is the early use of antibiotics,4 and their efficacy reduces by 7.5%–10% with each hour that antibiotics are delayed.5 Subsequently, once the disease process is established, antibiotics have little effect on treatment of fulminant sepsis.4 Other treatments, such as removal of cytokines, endotoxins and the involvement of OX40 in T-cell activation, are under investigation. Despite the complex process involved in the differential diagnosis of sepsis, a picture of multiorgan failure (MOF) slowly develops.6 Sepsis compromises respiratory function in the form of acute respiratory distress syndrome (ARDS) and compromises cardiac function (hypotension and lactic acidosis).7 Acute kidney injury (decreased urine output and increasing serum lactate levels) soon follows respiratory and cardiac compromise.8 The contemporary rationale for this failure of major organs is activation of the immune system originating from infective or traumatic cell damage at a whole body systemic, cellular level. This stimulates the inflammatory and anti-inflammatory cascades to aid the immune system to fight infection or repair tissue trauma.4

The removal of inflammatory/anti-inflammatory mediators has been identified as a potential therapy for sepsis in the belief that mediators could be removed from the circulation by the use of haemofiltration (HF).9 In the military context, it has been suggested that adsorption of mediators may be a treatment for ‘holding’ combat casualties suffering from sepsis until casualty evacuation recommences.10

The potential intervention

The removal of the mediator molecules from blood can be achieved by initial plasma filtration. Plasma is centrifuged from the blood, and mediators are removed by passing the plasma through an adsorption material filter before returning the plasma to the whole blood. Blood is then haemofiltered to remove urinary by-products and returned to the patient.11 Known as coupled plasma filtration adsorption (CPFA), it has been observed in small studies to reduce mortality12 and could provide an effective tool in the treatment of severe sepsis in addition to antibiotics and organ support.4

The specific focus on the removal of mediators has turned to unselective methods of adsorption rather than direct filtration.13 CPFA filters plasma and the inflammatory and anti-inflammatory mediators from blood; it then passes the plasma through a non-selective adsorbent filter to remove mediators.14 Plasma is returned to blood with a significantly reduced mediator loading, which, in turn, reduces systemic ‘stress’ on the body. Ronco et al 15 suggest that the reduction or prevention of mediator levels reaching peak concentration can prevent immunoparalysis. It is hypothesised that the removal of mediators and subsequent systemic stress improves homeostasis, reducing requirements for clinical support and polypharmacy, but there is ongoing uncertainty. A systematic review and meta-analysis were therefore performed to estimate the effects of CPFA on mortality in severe sepsis.

Materials and methods

The Cochrane Collaboration methodology was used to undertake, and the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) structure to report, a systematic review and meta-analysis to establish the effect of coupled plasma HF filtration (CPFA) in comparison with HF in patients suffering from sepsis. The primary outcome was 28-day mortality; secondary ones were mean arterial pressure (MAP), oxygen ratio and cytokine clearance. The review was not registered with PROSPERO.

The following electronic databases (1997–2013) were searched in December 2013: British Nursing Index, CINAHL, Cochrane, EMBASE, MEDLINE—EBSCO-Host, MEDLINE—ProQuest and National Institute for Health and Care Excellence (NICE). Hand searching of references of identified articles, narrative articles, supplements and editorials was used to identify further relevant studies; there were no language exclusions and foreign language articles were translated. Randomised controlled trials (RCT), prospective cohort studies (PCS), prospective crossover studies (PCOS) and pre-and-post studies (PPS) were included. Patients in the RCTs and PCS were treated with either HF or CPFA; patients in the PCOS and PPS were treated with both HF and CPFA. Abstracts were screened, and those not relevant to the review were excluded. Full text of articles potentially meeting inclusion criteria were obtained and independently assessed, with disagreements resolved by discussion.

Data extraction

For each study, data were abstracted using a standardised study pro forma, including the following study characteristics: countries and study sites, study dates, number of patients, randomisation, reported primary endpoint and timing of mortality assessment and secondary outcomes where relevant. Articles were assessed for quality of evidence using the Grading of Recommendations, Assessment, Development and Evaluation Working Group16 criteria. Risk of bias was assessed using the Critical Appraisal Skills Programme17 framework. A funnel plot was constructed to assess possible publication bias using RevMan V.5.1, the software used for preparing and maintaining Cochrane Reviews. To externally validate decisions made during study selection and data abstraction, 50% of articles were randomly selected for independent review, and a κ score calculated to assess strength of agreement; the κ Score was a perfect 1.0.

Synthesis of results

Meta-analysis was performed using RevMan V.5.1. To obtain statistical data, Comprehensive Meta-Analysis 2 was used. Baseline value, effect size and SD/CIs were calculated by generic point estimates using data from a raw scale to compute the effect size. A 95% CI was extracted from included articles, and a random effects model was used because not all the studies estimated all the same values.


The literature search is summarised in Figure 1.

Figure 1

Literature search summary.

Twenty eligible studies were identified from 52 articles screened. Following quality/bias assessment, 17 articles were included (n=441 patients, of whom 242 had CPFA and 199 HF). Fourteen studies evaluated the effect of CPFA compared with HF on mortality, 13 studies compared CPFA with HF on improving MAP, 12 measured the effect of CPFA versus HF in improving oxygen ratio and 12 papers evaluated CPFA versus HF on reducing mediator loads.

Study characteristics

There were three RCTs18–20, three PCOS,21–23 six PCS,24–28 four prospective PPS12 ,14 ,29 ,30 and one where the study design was unclear.31 All studies used plasma adsorption and six identified the type of adsorption material.14 ,18–20 ,24 One study evaluated the effect of CPFA on postoperative patients with diagnosed sepsis due to peritonitis,14 and two evaluated the effect of CPFA on sepsis caused by specific bacterial types.18 ,29

Risk of bias

A funnel plot was constructed (effect size vs SE) to assess for possible publication bias expressed by asymmetry and clustering (Figure 2).32 Complete papers were obtained and screened for construct bias using the Critical Appraisal Skills Programme frameworks.17 On conclusion of this process, the results were tabulated and conclusions drawn for analysis. Those papers that showed significant bias were removed and not included in the review; however, bias was found in the remaining 17 papers.

Figure 2

Funnel plot to assess possible bias.

The sensitivity bias/publication bias funnel plot analysis was calculated by RevMan V.5.1 on a risk ratio (RR) with 95% CI, based on the primary outcome measure (mortality). No overt bias (clusters or overt asymmetry) of statistical heterogeneity was observed in the I2 value or the χ2 test: I2=26% and the χ2=12.09.

Allocation and blinding

Random sequence generation was reported in only three studies.18–20 Blinding was not possible due to the nature of the treatment and the impracticality of blinding of ICU staff due to the small numbers involved.33 Allocation concealment was adequately reported in only one study.20

Primary outcome

Fourteen papers reported on the primary outcome of 28-day mortality. Thirteen studies used the measure of 28-day mortality to ascertain efficacy, whereas Suzuki et al 18 used a 14-day mortality measure. By statistically combining the data from those studies, the pooled effect favours CPFA (Figure 3). When Suzuki et al's 18 data were removed, the result still favoured CPFA (Figures 3 and 4). The effect size of the two RCTs was no different from that of non-randomised studies. The effect size of the different types of adsorbent filter used could not be calculated due to lack of information.

Figure 3

28-Day mortality pooled effect (RevMan V.1).

Figure 4

28-Day pooled mortality (CMA2). A = CPFA - B = HF

Secondary outcomes

Mean arterial pressure

Thirteen studies reported the effect of CPFA on MAP, which on average improved by 15 mm Hg with CPFA compared with no statistically significant improvement in MAP in the HF group. However, only five studies reported a mean to offer a comparative point estimate of effect; the pooled effects of studies favoured CPFA: OR 0.34 (95% CI 0.24 to 0.13) (Figure 5).

Figure 5

Pooled results for effect on mean arterial pressure (MAP). A = HF - B = CPFA

Oxygen ratio

Although 12 studies reported the effect of CPFA on oxygen ratio, only four presented data that could be analysed. In the CPFA group, oxygen ratio increased on average by 11 units (OR 0.908 (95% CI 0.695 to 1.18)) (Figure 6); oxygen ratio did not increase in HF patients.

Figure 6

Pooled results for effect on oxygen ratio mediators. A = HF - B = CPFA

Twelve papers reported the effect of CPFA on removal of cytokines; however, due to the variation in reported assay methods (different cytokines and using different methods of analysing cytokines), only five articles were analysed, all of which identified Toxicolor as the method of assay: OR 0.331 (95% CI 0.283 to 0.386) (Figure 7).

Figure 7

Pooled effects on mediators. A = HF - B = CPFA


The inclusion of such small underpowered studies in a systematic review/meta-analysis is subject to publication bias34; a funnel plot was used in order to analyse the data and address this limitation. Information was sought from authors of conference abstracts, which had not resulted in publication, but no responses were received.

The pooled effects are weighted on the size of the study; the greater the size the greater the value of that effect on the review, (Professor M Bland, personal communication). While this may be true for RCTs, it is a limiting factor for a SR/MA made up of heterogeneous methods. Weighting in such heterogeneous studies should include a ratio of weighting to accommodate for the quality of the study method.


While positive effects on mortality, MAP, oxygenation, and (despite the poor dataset), mediator removal were reported, the quality of published studies, and the small number of patients, preclude firm conclusions being drawn. Large randomised trials are required to address this important treatment uncertainty.


Professor Lui Forni, University of Surrey and Surrey Perioperative Anaesthesia and Critical Care Collaborative Research Group, Gp Capt Di Lamb PhD, Professor of Nursing, Academic Department of Military Nursing, Royal Centre for Defence Medicine (Academia and Research). Lt Col Lizzy Bernthal PhD, QARANC, Senior Research Fellow, Academic Department of Military Nursing, Royal Centre for Defence Medicine (Academia and Research).


View Abstract


  • Twitter Follow Tom Quinn at @proftomquinn

  • Contributors IH: original author. TQ: author and editor. SJ: reviewed data collection and analysis.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; internally peer reviewed.