NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Clinical Guideline Centre (UK). Chronic Heart Failure: National Clinical Guideline for Diagnosis and Management in Primary and Secondary Care: Partial Update [Internet]. London: Royal College of Physicians (UK); 2010 Aug. (NICE Clinical Guidelines, No. 108.)

Cover of Chronic Heart Failure

Chronic Heart Failure: National Clinical Guideline for Diagnosis and Management in Primary and Secondary Care: Partial Update [Internet].

Show details

Appendix HCost-Effectiveness Analysis

Cost-effectiveness analysis of serial measurement of circulating natriuretic peptide concentration in patients with heart failure

1. Background

Brain natriuretic peptide (BNP) and its aminoterminal portion (N-BNP) are secreted primarily from the left ventricle in response to changes in left ventricular wall stretch1. These agents are neurohormonal predictors of left-ventricular function and prognosis2, 3, 4, 5. The diagnostic and prognostic value of natriuretic peptide plasma level in heart failure was supported by many studies6, 7, 8, 9, 10,11. It was also proven that most drugs used to treat heart failure significantly reduce natriuretic peptide level12,13,14,15,16,17.

Treatment optimization for patients with heart failure is based on physician assessment and patient tolerance. Circulating natriuretic peptide concentration can be reduced by intensification of drug therapy in heart failure, and monitoring plasma natriuretic peptide level has been proposed for optimizing medical treatment. Four randomised clinical trials were published comparing the management of patients’ medical treatment according to natriuretic peptide concentration versus clinical assessment in secondary care and/or usual care in the community18,19,20,21. These clinical trials reported that serial measurement of natriuretic peptide concentration improved outcomes compared to clinical assessment or usual care.

In England and Wales, natriuretic peptide measurement is available but its use as a monitoring tool is not widespread. National implementation might significantly affect resource use in the NHS. One cost-effectiveness analysis was published assessing the management of medical treatment in chronic heart failure using BNP measurement compared to clinical assessment22. This analysis was based on one clinical trial18 and it showed that BNP monitoring was cost-effective. However, this analysis was developed from a US perspective, and the generalisation of these results to a UK context is questionable. Furthermore, there is now considerably more trial evidence. Therefore, we undertook an original cost-effectiveness analysis from a UK NHS and personal social services perspective.

2. Objective

The objective of this economic analysis was to assess the cost-effectiveness of three alternative strategies:

  • serial measurement in secondary care of circulating natriuretic peptide concentration for optimizing medical therapy
  • clinical assessment in secondary care
  • usual care in the community

for patients in England and Wales with

  1. chronic heart failure (CHF), or
  2. CHF and left ventricular systolic dysfunction (LVSD).

3. Model

In a systematic clinical review [see Section 7.1.2 of Full Guideline (2010)], four clinical trials were identified assessing serial measurement of natriuretic peptide concentration for optimizing the medical therapy in CHF (Troughton 200018, Jourdain 200720, Pfisterer 200919, Lainchbury 201021)b. Trougton 200018, Jourdain 200720, and Pfisterer 200919 compared serial measurement in secondary care of natriuretic peptide concentration and clinical assessment in secondary care. Lainchbury 201021 compared natriuretic peptide measurement in secondary care, clinical assessment in secondary care, and usual care in the community.

The Trougton 200018, Jourdain 200720, and Pfisterer 200919 clinical trials were conducted in patients with CHF and LVSD. Lainchbury 2010 clinical trial21 was conducted on patients with CHF of any causes. Hence, outcomes of the three clinical trials on patients with LVSD 18, 20, 19 were meta-analysed for use in this economic analysis, and outcomes from the Lainchbury clinical trial21 were utilized independently. Furthermore, age subgroups were assessed in Pfisterer19 (<75 years/≥75 years) and Lainchbury21 (≤75 years/>75 years), and cost-effectiveness analyses were therefore conducted for these subgroups.

The same mortality rate and yearly cost per patient were assumed for each intervention after the trial periods (Section 4.1.2 and 5.6). A lifetime horizon was used when the number of patients who were alive differed between the compared cohorts at the end of the trial follow-up. When the same number of patient was alive in each trial arm at the end of the trial, the trial period was used as the model time horizon. It was judged that the same number of patient were alive in the three compared cohorts at the end of Lainchbury main analysis, and between the clinical assessment and the usual care cohorts in Lainchbury age-subgroup analyses (≤75 years/>75 years) (Table 1) 21. Therefore, cost-effectiveness assessments were conducted on these analyses on a three-year time horizon. In addition, for Lainchbury21 age subgroups, cost-effectiveness assessments were conducted on a lifetime horizon as a higher proportion of patients were alive at the end of the trial in natriuretic peptide cohorts in comparison to clinical assessment or usual care.

Table 1. Mortality (all-cause) - Risk ratios (95% confidence intervals).

Table 1

Mortality (all-cause) - Risk ratios (95% confidence intervals).

Cost-effectiveness analyses were developed from an England and Wales NHS perspective; the health outcome considered was Quality-Adjusted Life Year (QALY), and an annual discount rate of 3.5% was applied to both costs and health outcomes incurred after one year.

4. Quality-Adjusted Life Year

Quality-Adjusted Life Years (QALYs) are calculated by multiplying the patients’ life expectancy (life years) by a utility score (a quality of life measure on a 0–1 scale).

4.1. Mortality

Within-trial mortality estimates were taken from the clinical trials themselves. Patients’ mortality post-trial was assumed the same for each compared cohort in all the analyses. Post-trial mortality estimates were taken from the UK-based study conducted on patients with heart failure by Guili 200524.

4.1.1. Mortality within-trial

Two techniques were used to estimate life years for the within-trial periods. When survival curves were available, life years where calculated as the area under the survival curve. Alternatively, risk ratios at the end of trials were used assuming deaths occurred evenly over the trial follow-up period.

The area under the curve was calculated for assessments developed from Lainchbury21 and Pfisterer19. Table 2 shows life years calculated from survival curves. As explained in Section 6, the Pfisterer trial19 (all ages) was modelled independently as a sensitivity analysis.

Table 2. Within-trial life years calculated as the area under the survival curve.

Table 2

Within-trial life years calculated as the area under the survival curve.

Risk ratios were used to calculate life years in the cost-effectiveness assessment based on trials conducted on patients with CHF and LVSD (Troughton18, Jourdain20, and Pfisterer19). The meta-analysed risk ratio (Table 3) was applied at 18 months (Pfisterer trial19 follow-upc). The baseline risk used was the death risk from Pfisterer19, the largest trial, in the clinical assessment cohort at 18 months (Table 3). In addition, we modelled as part of the sensitivity analysis (Section 6) Jourdain20 and Pfisterer19 independently (Table 3).

Table 3. Mortality all-cause.

Table 3

Mortality all-cause.

4.1.2. Mortality post-trial

Giuli 200524 reported outcomes from an observational study using the General Practice Research Database in the UK. This study aimed to determine the incidence and prognosis of heart failure (HF) diagnosed by general practitioners. Incident cases of HF in 1991 were selected and followed for three-years. 686,884 patients 45 years and older were classified as definite HF, possible HF, or a prescription of diuretics without a diagnosis of HF. 6478 patients were classified as definite HFd, and outcomes from this subgroup were considered relevant for this cost-effectiveness analysis.

The mean survival time for definite HF was 23.8 months (95% CI 23.4–24.1), 22.9 months in men and 24.5 months in women ( p<0.001). The median survival was 30.8 and 36.5 months in men and women respectively. Sex- and age-group standardised mortality ratios (SMR) were reported. Each SMR was the ratio of the cumulative probability of dying in the study population to the cumulative probability of dying in an age and sex matched sample from the general population in England and Wales. We adjusted the SMRs to account for the effect on survival of ACEI and BB using data from meta-analyses by Flather 200025 for ACEI and Shibata 200126 for BB assuming no interaction between the two drugse. Table 4 presents both the unadjusted SMR estimates from Guilli 200524 (untreated), and our adjusted estimates, which we used in our cost-effectiveness analysis.

Table 4. Standardised mortality ratios (definite HF vs general population).

Table 4

Standardised mortality ratios (definite HF vs general population).

We estimated life expectancy beyond the trial follow-up using the official life tables for England and Wales27 but adjusting the mortality using the CHF-specific SMRs (Table 4). The life expectancies were based on the mean age at baseline from the trials and were at first calculated for men and women separately18, 19, 20, 21. Then, we calculated the average life expectancy for both sexes using the male/female ratio at baseline in clinical trials18, 19, 20, 21. Table 5 presents the life expectancies from trial baseline considered for our economic analysis.

Table 5. Life expectancy from baseline.

Table 5

Life expectancy from baseline.

4.2. Utility scores

The four clinical trials18, 19, 20, 21 did not report utility scores. There were some assessments of patients’ health-related quality of life (HRQoL) and functional capacitiesf, but these could not be used to estimate utility.

Gohler 200928 reported mean utility scores stratified by NYHA class for patients with CHFg. They used EuroQol 5D (EQ-5D) data collected from the EPHESUS trial29 (multi-centre and multi-national trial), which assessed the addition of eplerenone to optimal medical treatment in patients with CHF and LVSD post myocardial infarction. During the EPHESUS trial29, EQ-5D data were collected from a subsample of 1628 patients at baseline, three, six, 12, and 18 months. Gohler 200928 estimated utilities using all except the baseline data to mitigate the effect of acute myocardial infarction on the EQ-5D score (Table 6).

Table 6. Utility score – Patients with Chronic heart Failure.

Table 6

Utility score – Patients with Chronic heart Failure.

We estimated average utility scores for each of our trials by weighting each of the utility scores in Table 6 with the proportion of patients in each NHYA class at trial baseline (Table 7) 18, 19, 20, 21. In the absence of evidence to the contrary, we assumed that mean utility scores stayed constant over time and were the same for each intervention.

Table 7. Utility scores used in the economic analysis.

Table 7

Utility scores used in the economic analysis.

5. Resource use and cost

Resource use was taken from the clinical trials and was combined with standard UK unit costs. Resource use components considered were hospitalisation, drug usage, outpatient visits, natriuretic peptide assessment, and biochemistry testing to assess renal function. For the post-trial period, a yearly cost per patient was applied.

5.1. Hospitalisation

To estimate hospitalisation costs, we used the risk ratio from the final trial follow-up and we assumed admissions occurred evenly over the follow-up period. The hospitalisation risk for the clinical assessment cohort was used as the baseline risk. For the analysis conducted on patients with CHF and LVSD (based on Troughton18, Pfisterer19, and Jourdain20), we applied the meta-analysed risk ratio to the baseline risk at 18 months in the Pfisterer19 trialh. Table 8 details the trial hospitalisation data and the probabilities used in this cost-effectiveness analysis.

Table 8. Hospitalisation.

Table 8

Hospitalisation.

The hospitalisation cost per hospital admission was calculated from reported figures of the NHS reference cost30 databasei. This cost was estimated to be £1,725 and was combined with the probabilities in Table 8 to give the hospitalisation cost.

5.2. Drug usage

The change in drug usage was calculated for all clinical trials (Lainchbury21, Jourdain20, Pfisterer19, and Troughton18). In the cost-effectiveness assessment for patients with CHF and LVSD (based on Jourdain20, Pfisterer19, and Troughton18), the Pfisterer19 drug usage was used for the base case. The drug usage from the other trials was used in sensitivity analyses, to see if the source of this component can affect the results of the analysis (Section 6).

The Lainchbury21 drug usage was reported for the main analysis only (all ages). In the absence of better evidence, we assumed in our cost-effectiveness analysis that these data also applied to the age subgroups. The Pfisterer21 drug usage was calculated separately for the main analysis (all ages) and for the age subgroups.

5.2.1. Lainchbury

For the Lainchbury main analysis, mean daily drug doses per patient were reported at every follow-up assessment for furosemide (loop diuretic), enalapril (ACEI), metoprololj (BB), and spironolactone (Table 9).

Table 9. Lainchbury drug usage (mg/day).

Table 9

Lainchbury drug usage (mg/day).

We assumed that drug dosages changed at the mid-point between follow-ups. The usage at 24 months was assumed to stay constant up to 36 months (end of trial21). We combined these trial data with drug unit costs31; Table 10 presents costs of drug treatment for compared cohorts.

Table 10. Lainchbury drug treatment cost per patient.

Table 10

Lainchbury drug treatment cost per patient.

5.2.2. Pfisterer

Pfisterer19 presented at baseline the mean percentage of target dose per patient for ACEI/ARB, BB, and loop diuretic. In the absence of better data, these figures at baseline were assumed the same for the main analysis (all ages) and for age subgroups (<75 years, ≥75 years). Changes in drug usage were reported by age subgroups only, in percentage of target doses, for ACEI/ARB and BB. For the cost-effectiveness assessment based on the main analysis (all ages), changes in ACEI/ARB and BB usage were assumed to be the unweighted average of the reported changes for age subgroups. For jointly reported figures for ACEI and ARB, we costed the use of enalapril (ACEI), considering a target dose of 10mg b.i.d.k Carvedilol was costed as BB, considering a target dose of 25mg b.i.d.k No change in usage was reported for loop diuretic and this treatment was excluded from cost-effectiveness assessmentsl. Table 11 presents data for ACEI and BB used for cost-effectiveness assessments.

Table 11. Pfisterer drug usage.

Table 11

Pfisterer drug usage.

The number of patients taking spironolactone or eplerenone at baseline and at the end of interventions (6 months) was presented for the Pfisterer main analysis (all ages). In the absence of better data, these figures were also applied to the age-subgroup analyses. We assumed patients were taking spironolactone or eplerenone at a dose of 25mg/day. Table 12 presents drug usage data for spironolactone and eplerenone.

Table 12. Pfisterer drug usage.

Table 12

Pfisterer drug usage.

We assumed that drug treatments changed at three months (mid-point between the baseline and the end of interventions), and we assumed that the drug usage at six months stayed constant up to the end of follow-up (18 months)19. Table 13 presents costs of drug treatment for the compared cohorts.

Table 13. Pfisterer drug treatment cost per patient.

Table 13

Pfisterer drug treatment cost per patient.

5.2.3. Jourdain

Jourdain20 reported changes in drug usage for ACEI/ARB, BB, and furosemide (loop diuretic). The mean percentage of daily target dose per patient was reported at baseline and 3 months (end of the trial intervention) for ACEI/ARB and BB. The mean daily dose per patient at baseline and 3 months was reported for furosemide. For jointly reported figures for ACEI and ARB, we costed the use of enalapril (ACEI), assuming a target dose of 10mg b.i.d.k Carvedilol was costed as the BB, assuming a target dose of 25mg b.i.d.k Table 14 presents the drug usage data from Jourdain20.

Table 14. Jourdain drug usage.

Table 14

Jourdain drug usage.

The change in drug usage was assumed at 1.5 months (mid-point between the baseline and the end of interventions). In the absence of better data, we kept constant the drug usage at three months up to the end of Jourdain follow-up (15 months)20 for the analysis based on this trial alone (sensitivity analysis – Section 6), or up to 18 months, the Pfisterer19 follow-up periodm, when applying the Jourdain drug usage to the analysis developed on patients with CHF and LVSD (sensitivity analysis – Section 6). Table 15 presents costs of drug treatment for compared cohorts.

Table 15. Jourdain drug treatment cost per patient.

Table 15

Jourdain drug treatment cost per patient.

5.2.4. Troughton

Troughton18 reported the mean dose per patient at baseline and the mean dose increase per patient during the intervention period (6 months) for enalapril (ACEI) and furosemide (loop diuretic). The number of patients taking spironolactone and a BB at baseline and six months was also reported. We assumed that spironolactone was taken at a dose of 25mg/day, and that the BB carvedilol was taken at a dose of 25mg bd. Table 16 details the drug usage from Troughton18.

Table 16. Troughton drug usage.

Table 16

Troughton drug usage.

The change in drug usage was assumed at three months (mid-point between baseline and the end of trial intervention)18. We kept constant the drug usage at six months up to 18 months, which was the follow-up time of Pfisterer19 trialm. Table 17 presents costs of drug treatment for the compared cohorts.

Table 17. Troughton drug treatment cost per patient.

Table 17

Troughton drug treatment cost per patient.

Drug usages were costed using drug unit costs proposed by the British National Formulary31 (Table 18).

Table 18. Drug prices.

Table 18

Drug prices.

5.3. Outpatient visits

Table 19 presents numbers of outpatient visits attended from each of the four clinical trials. Troughton18 was the only clinical trial reporting additional (unplanned) outpatient visits. In the absence of better data, we assumed the Troughton figures for additional visits in our analyses of the Jourdain and Pfisterer trials. For Lainchbury, we assumed Troughton figures of additional visits for natriuretic peptide and clinical assessment cohorts, and no additional visit was assumed for the usual care cohort.

Table 19. Outpatient visits.

Table 19

Outpatient visits.

Natriuretic peptide and clinical assessment interventions were offered in secondary care at a specialist level in every clinical trial. The outpatient visit cost for these cohorts was calculated using figures reported by the National reference cost30 databasen, and was estimated to be £98 per visit. In the Lainchbury usual care cohort, it was conservatively assumed that all attendances were with the general practitioner. The mean cost per GP visit in the community was estimated nationally to be £5234.

5.4. Natriuretic peptide assessment

Natriuretic peptide assessments were undertaken at every outpatient visit in the natriuretic peptide cohort. There is no national price for this test in England and Wales. The tariff price at St George’s Healthcare Trust (London) is £27.71 for NT-proBNP testingo. This cost was used for base-case cost-effectiveness assessments and added to the cost of an outpatient visit (Section 5.3)p. To allow for a potentially lower cost for natriuretic peptide testing, for example if this test is made available to a large number of patients, we used in the sensitivity analysis a cost of £20 (Section 6).

5.5. Biochemistry testing

When initiating or modifying dosages of ACEI, diuretic, and spironolactone/eplerenone, biochemistry testing for renal function is current practice. Numbers of treatment modifications per cohort were reported by Jourdain20 for these drugs (Table 20). Pfisterer19 reported the number of patients per cohort adding spironolactone/eplerenone to their drug therapy during interventions (none were taking spironolactone/eplerenone at baseline) (Table 20). We calculated probabilities of treatment modifications for natriuretic peptide and clinical assessment cohorts using data from Jourdain20 for ACEI, and diuretic, and pooled data from Jourdain20 and Pfisterer19 for spironolacone/eplerenone (Table 20). In the absence of data for Lainchbury usual care cohort, we assumed no biochemistry testing for this group of patients.

Table 20. Treatment modifications.

Table 20

Treatment modifications.

The probability of treatment modification was multiplied by the average cost of a biochemistry test: £1.34 from the NHS Reference costs30. The cost of biochemistry testing may have been overestimated for natriuretic peptide and clinical assessment cohorts, as multiple treatment modifications may occur during a single physician visit. However, since the cost of biochemistry testing is so small, the impact on the results of our cost-effectiveness analysis is minimal.

5.6. Post-trial cost

Stewart 200235 published a cost-of-illness analysis of heart failure developed from a UK NHS perspective. Cost components incorporated in the analysis were hospitalisation, hospital-based outpatient consultations, GP consultations, drug treatment, and nursing-home care. The yearly cost per patient was estimated in 2000 to be £896q. Using the prices index for hospital and community health services34, we estimated this cost in 2008 GBP to be £1,171 per patient per year. This yearly cost per patient was used in the post-trial period of the model was assumed the same for the different cohorts.

6. Sensitivity analysis

Sensitivity analyses were performed to assess the robustness of the cost-effectiveness results to plausible variations in model parameters. First, for the cost-effectiveness assessment conducted on patients with CHF and LVSD, the Pfisterer19 drug usage was used for the base case, and drug usages from Jourdain20 and Troughton18 (Section 5.2) were applied to sensitivity analyses.

Secondly, Jourdain20 and Pfisterer19 clinical trials were modelled independently in addition to the assessment combining outcomes from Pfisterer19, Jourdain20, and Troughton18, because of some inconsistencies in outcomesr. Troughton18 was not modelled independently since it was small and did not report all-cause mortalitys.

Furthermore, as discussed in Section 3, the same number of patients was alive in the three compared cohorts at the end of Lainchbury main analysis, and between the clinical assessment and the usual care cohorts in Lainchbury age-subgroup analyses (≤75 years/>75 years) (Table 1)21. Thereby, the cost-effectiveness assessment from Lainchbury21 main analysis was conducted on a three-year time horizon, and cost-effectiveness assessments from Lainchbury21 age-subgroup analyses were conducted on both a three-year and a lifetime horizons. Moreover, cost-effectiveness assessments conducted on patients with CHF and LVSD were developed on a lifetime horizon in the base-case analysis. These cost-effectiveness assessments were based on trial follow-ups shorter than three years (18 months19 and 15 months20). Considering that mortality ratios in natriuretic peptide and clinical assessment cohorts for all-age analyses might be the same at three years as in Lainchbury21 main analysis, we conducted additional analyses on patients with CHF and LVSD on a three-year time horizont.

Finally, as discussed in Section 5.4, we used in the sensitivity analysis a cost of £20 for natriuretic peptide testing in all cost-effectiveness analyses in addition to the £27.71 used in the base case.

7. Probabilistic analysis

This economic analysis presents probabilistic results. A probabilistic analysis applies probability distributions to each model parameter and therefore allows us to calculate a distribution for the results of the cost-effectiveness analysis, equivalent to a confidence interval. A gamma distribution (bounded at 0) was applied to cost estimates and to standardized mortality ratios. A beta distribution (bounded between 0 and 1) was applied to utility scores and probabilities. Finally, a lognormal distribution (bounded at 0) was applied to risk ratios, mean drug dosageu and mean number of outpatient visits (refer to Table 25 on Section 11). The results of each analysis (base-case analyses and sensitivity analyses) were re-calculated 5000 times, with all the model parameters set simultaneously, selected at random from the respective parameter distribution. We present the results in terms of the mean of the 5000 computed simulations.

8. Results

This economic analysis assessed two populations of patients: patients with CHF and LVSD; and patients with heart failure of any cause. For these two populations, age subgroups were also assessed (Pfisterer <75 years, ≥75 years; Lainchbury ≤75 years, >75 years).

8.1. Patients with CHF and LVSD

Table 21 presents the breakdown of resource use components, life years, and QALYs for the base-case cost-effectiveness analysis developed on patients with CHF and LVSD based on the Pfisterer19, Jourdain20, and Troughton18 trials. Table 22 presents cost-effectiveness results for the base-case analysis, subgroup analyses, and sensitivity analysis in this population. Results show that serial measurement of natriuretic peptide concentration in secondary care is clearly cost-effective compared to clinical assessment in secondary care, for the base-case population and both age subgroups (<75 years, ≥75 years). The probability of natriuretic peptide being cost-effective was high (98% for the base case, 99% for <75 years, and 68% for ≥75 years). The conclusion was the same in all the sensitivity analyses. In the sensitivity analysis based on Jourdain20 with a three-year time horizon, the natriuretic peptide option was actually cost-saving compared to clinical assessment.

Table 21. Cost and QALY results: Patients with CHF and LVSD (lifetime horizon).

Table 21

Cost and QALY results: Patients with CHF and LVSD (lifetime horizon).

Table 22. Cost-effectiveness results: Patients with CHF and LVSD (natriuretic peptide vs clinical assessment).

Table 22

Cost-effectiveness results: Patients with CHF and LVSD (natriuretic peptide vs clinical assessment).

8.2. Patients with CHF due to any cause

The population assessed in Lainchbury21 was patients with CHF due to any cause. Based on Lainchbury21, we assessed the cost-effectiveness of serial measurement in secondary care of natriuretic peptide concentration compared to a) clinical assessment in secondary care and to b) usual care in the community. In addition to the base-case cost-effectiveness assessment developed from the main Lainchbury results, age subgroups analyses were also conducted (<75 years, ≥75 years)21.

Table 23 presents a breakdown of cost components, life years, and QALYs for the base-case cost-effectiveness analysis developed from Lainchbury21. Table 24 shows results of this cost-effectiveness analysis modelled on a three-year time horizon (Section 3). Comparing an intervention with the next best alternative (Figure 1), and applying a threshold of £20,000 per QALY gained, clinical assessment is cost-effective compared to usual care (ICER = £7,188/QALY) and natriuretic peptide is cost-effective compared to clinical assessment (ICER = £11,861/QALY). Serial measurement of natriuretic peptide is therefore the preferred option from a cost-effectiveness perspective.

Table 23. Cost and QALY results: Patients with CHF of any cause - Lainchbury (3 years time horizon).

Table 23

Cost and QALY results: Patients with CHF of any cause - Lainchbury (3 years time horizon).

Table 24. Cost-effectiveness: CHF of any cause - Lainchbury.

Table 24

Cost-effectiveness: CHF of any cause - Lainchbury.

Figure 1. Cost-effectiveness results (CHF any cause; base case).

Figure 1

Cost-effectiveness results (CHF any cause; base case).

For the age-subgroup cost-effectiveness assessment conducted on patients 75 years old and younger and developed on three-year and lifetime horizons (Section 3), the diagram of the cost-effectiveness plane (Figure 2) shows that clinical assessment is ruled out due to ‘extended dominance’. Extended dominance exists when an option is less effective and more costly than a linear combination of two other strategies. The results show that serial measurement in secondary care of natriuretic peptide is highly cost-effective compared to usual care in the community for patients with CHF 75 years old and younger (Table 24).

Figure 2. Cost-effectiveness results (CHF any cause; age subgroups).

Figure 2

Cost-effectiveness results (CHF any cause; age subgroups).

For the age-subgroup cost-effectiveness assessment conducted on patients older than 75 years and developed on three-year and lifetime horizons (Section 3), the natriuretic peptide option is dominated by usual care (usual care is more effective and less costly – Figure 2). However, clinical assessment is cost-effective compared to usual care (Table 24). Therefore, clinical assessment in secondary care is the preferred options for patients with CHF older than 75 years.

Finally, the results of all analyses stayed the same when using a cost of £20 for natriuretic peptide testing (instead of £27 – Section 5.4).

9. Discussion

We assessed the use of serial measurement in secondary care of natriuretic peptide for optimizing medical therapy in patients admitted to hospital because of chronic heart failure, compared to both clinical assessment in secondary care and to usual care in the community:

  • Clinical assessment was more costly than usual care
  • Clinical assessment was more effective and cost-effective compared to usual care
  • Natriuretic peptide monitoring was more costly than clinical assessment (with exception of the analysis based on Jourdain20 and the one based on Lainchbury21 >75)
  • Natriuretic peptide monitoring was more effective and cost-effective compared to clinical assessment (with exception of the analysis based on Lainchbury21 >75)
  • Conclusions stayed consistent for age subgroups for patients with CHF and LVSD
  • Clinical assessment was the preferred option in patients older than 75 years with CHF due to any cause
  • Results were robust to sensitivity analyses

At the end of the Lainchbury trial21, the same number of patients was alive in the three compared cohorts. In the base-case cost-effectiveness analysis based on Lainchbury21 (patient with CHF due to any cause), the natriuretic peptide option being cost-effective relates to the calculation of life years using survival curves, which is more precise than using end-of-trial risk ratios. However, where we used survival curves to calculate life years, sampling error was not accounted for and uncertainty was underestimated. Nevertheless, for the analysis of patients with CHF and LVSD, which did not use this approach, the probability that natriuretic peptide monitoring is cost-effective was still convincingly high (98.3%).

Additional outpatient visits for up titrating medical therapy were reported by Troughton18 only and were applied to all cost-effectiveness analyses for natriuretic peptide and clinical assessment cohorts. Troughton18 was conducted before beta blockers were commonly used in heart failure and this may mean that we have under-estimated the additional outpatient visits associated with natriuretic peptide monitoring and therefore under-estimated the cost-effectiveness ratio.

In cost-effectiveness assessments of Lainchbury’s age subgroups, using lifetime or three-year time horizons did not change conclusions. However, when comparing clinical assessment and usual care in patients older than 75 years, the probability of clinical assessment being cost-effective compared to usual care was 50% on a lifetime horizon and 87% on a three-year time horizon. As the same number of patients were alive at the end of Lainchbury trial21 (3 years) in usual care and clinical assessment cohorts (in patients older than 75 years), the three-year time horizon results with the probability of cost-effectiveness of 87% are more relevant.

Results from cost-effectiveness assessments conducted on patients 75 years and older differed using outcomes from Lainchbury21 (>75) or from Pfisterer19 (≥75). The natriuretic peptide intervention improved survival in Pfisterer19 and decreased it in Lainchbury21 (compared to clinical assessment). It might be because patients with heart failure and preserved ejection fraction (HFPEF) were included in Lainchbury21 and excluded in Pfisterer19, and drug treatments in CHF were not shown to be as effective in HFPEF as they were in CHF with LVSD. The GDG also postulated that interventions in older CHF patients driven by raised natriuretic peptide can also increase the risk of renal impairment, thus adding to the potential risk of the NP-guided strategy in this age group.

Results presented are related to this population of patients, and may not be applied to patients excluded from clinical trials on which we based our cost-effectiveness analysis. The use of natriuretic peptide guided intervention in general practices was not assessed in clinical trials and no conclusion can be drawn. Considering the influence of the outpatient visit cost in the Lainchbury cost-effectiveness analyses, it might be advantageous to implement serial measurement of natriuretic peptide concentration for optimizing CHF medical therapy in general practices. Additional research is needed.

10. Conclusion

The optimization of drug therapy in chronic heart failure using serial measurement in secondary care of natriuretic peptide concentration is cost-effective compared to clinical assessment in secondary care and to usual care in the community. However, the use of natriuretic peptide measurement in patients older than 75 years may be harmful and not cost-effective, which suggests that careful patient selection is important. However, for patients older than 75 years, the optimization of drug therapy in chronic heart failure by clinical assessment in secondary care without natriuretic peptide monitoring was still cost-effective compared to usual care in the community.

11. Parameters used in probabilistic analyses

Table 25Parameters used in probabilistic analyses

Description of variableMean valueProbability distributionParametersSource
Lainchbury
Mortality risk ratio
Lainchbury (all ages) NP vs Clinic1.00lognormal95% CI = 0.7; 1.43Lainchbury21
Lainchbury (all ages) UC vs Clinic0.99lognormal95% CI = 0.69; 1.42Lainchbury21
Lainchbury (≤75 yrs) NP vs Clinic0.50lognormal95% CI = 0.24; 1.03Lainchbury21
Lainchbury (≤75 yrs) UC vs Clinic1.01lognormal95% CI = 0.59; 1.73Lainchbury21
Lainchbury (>75 yrs) NP vs Clinic1.41lognormal95% CI = 0.93; 2.14Lainchbury21
Lainchbury (>75 yrs) UC vs Clinic0.99lognormal95% CI = 0.61; 1.61Lainchbury21
Mortality baseline risk
Lainchbury (all ages)0.33Betaα = 40; β = 81Lainchbury21; clinic cohort
Lainchbury (≤75 years)0.31Betaα = 17; β = 38Lainchbury21; clinic cohort
Lainchbury (>75 years)0.35Betaα = 23; β = 43Lainchbury21; clinic cohort
Hospitalisation for heart failure risk ratio
Lainchbury all ages - NP vs Clinic0.90lognormal95% CI = 0.65; 1.24Lainchbury21
Lainchbury ≤75 - NP vs Clinic0.73lognormal95% CI = 0.44; 1.23Lainchbury21
Lainchbury >75 - NP vs Clinic1.05lognormal95% CI = 0.7; 1.57Lainchbury21
Lainchbury all ages - UC vs Clinic0.83lognormal95% CI = 0.6; 1.15Lainchbury21
Lainchbury ≤75 - UC vs Clinic0.90lognormal95% CI = 0.57; 1.42Lainchbury21
Lainchbury >75 - UC vs Clinic0.76lognormal95% CI = 0.47; 1.23Lainchbury21
Hospitalisation for heart failure; Baseline risk
Lainchbury all ages0.40betaα = 49; β = 72Lainchbury21; clinic cohort
Lainchbury ≤75 yrs0.40betaα = 22; β = 33Lainchbury21; clinic cohort
Lainchbury >75 yrs0.41betaα = 27; β = 39Lainchbury21; clinic cohort
Drug usage (mg)
Furosemide
NP baseline128lognormalSE = 2.09Lainchbury21
Clinic baseline149lognormalSE = 2.09Lainchbury21
UC baseline124lognormalSE = 1.99Lainchbury21
NP 3 months138lognormalSE = 1.82Lainchbury21
Clinic 3 months144lognormalSE = 1.91Lainchbury21
UC 3 months121lognormalSE = 1.9Lainchbury21
NP 6 months140lognormalSE = 2Lainchbury21
Clinic 6 months134lognormalSE = 1.91Lainchbury21
UC 6 months119lognormalSE = 1.9Lainchbury21
NP 12 months182lognormalSE = 2Lainchbury21
Clinic 12 months166lognormalSE = 2.09Lainchbury21
UC 12 months123lognormalSE = 1.99Lainchbury21
NP 24 months200lognormalSE = 2.45Lainchbury21
Clinic 24 months197lognormalSE = 2.55Lainchbury21
UC 24 months140lognormalSE = 2.26Lainchbury21
Enalapril
NP baseline12.7lognormalSE = 0.55Lainchbury21
Clinic baseline13.3lognormalSE = 0.55Lainchbury21
UC baseline10.3lognormalSE = 0.54Lainchbury21
NP 3 months13.0lognormalSE = 0.55Lainchbury21
Clinic 3 months14.7lognormalSE = 0.55Lainchbury21
UC 3 months11.3lognormalSE = 0.54Lainchbury21
NP 6 months13.3lognormalSE = 0.55Lainchbury21
Clinic 6 months14.6lognormalSE = 0.55Lainchbury21
UC 6 months11.0lognormalSE = 0.54Lainchbury21
NP 12 months13.1lognormalSE = 0.55Lainchbury21
Clinic 12 months14.2lognormalSE = 0.55Lainchbury21
UC 12 months11.0lognormalSE = 0.54Lainchbury21
NP 24 months12.4lognormalSE = 0.64Lainchbury21
Clinic 24 months14.0lognormalSE = 0.64Lainchbury21
UC 24 months10.8lognormalSE = 0.54Lainchbury21
Sprionolactone
NP baseline20lognormalSE = 0.55Lainchbury21
Clinic baseline21lognormalSE = 0.55Lainchbury21
UC baseline20lognormalSE = 0.18Lainchbury21
NP 3 months22lognormalSE = 0.36Lainchbury21
Clinic 3 months22lognormalSE = 0.45Lainchbury21
UC 3 months20lognormalSE = 0.18Lainchbury21
NP 6 months22lognormalSE = 0.36Lainchbury21
Clinic 6 months24lognormalSE = 0.45Lainchbury21
UC 6 months21lognormalSE = 0.18Lainchbury21
NP 12 months20lognormalSE = 0.45Lainchbury21
Clinic 12 months23lognormalSE = 0.45Lainchbury21
UC 12 months21lognormalSE = 0.18Lainchbury21
NP 24 months16lognormalSE = 0.64Lainchbury21
Clinic 24 months20lognormalSE = 0.55Lainchbury21
UC 24 months21lognormalSE = 0.27Lainchbury21
Sprionolactone
NP baseline76lognormalSE = 11Lainchbury21
Clinic baseline80lognormalSE = 11Lainchbury21
UC baseline73lognormalSE = 10Lainchbury21
NP 3 months83lognormalSE = 9Lainchbury21
Clinic 3 months91lognormalSE = 9Lainchbury21
UC 3 months74lognormalSE = 9Lainchbury21
NP 6 months95lognormalSE = 9Lainchbury21
Clinic 6 months95lognormalSE = 9Lainchbury21
UC 6 months75lognormalSE = 9Lainchbury21
NP 12 months95lognormalSE = 10Lainchbury21
Clinic 12 months99lognormalSE = 10Lainchbury21
UC 12 months73lognormalSE = 10Lainchbury21
NP 24 months94lognormalSE = 11Lainchbury21
Clinic 24 months99lognormalSE = 12Lainchbury21
UC 24 months72lognormalSE = 10Lainchbury21
Pfisterer
Mortality risk ratio
Pfisterer (all ages)0.72lognormal95% CI = 0.5; 1.04Pfisterer19
Pfisterer (<75 yrs)0.47lognormal95% CI = 0.24; 0.92Pfisterer19
Pfisterer (≥75 yrs)0.91lognormal95% CI = 0.61; 1.37Pfisterer19
Mortality baseline risk
Pfisterer (all ages)0.22Betaα = 55; β = 193Pfisterer19; clinic cohort
Pfisterer (<75 years)0.22Betaα = 22; β = 80Pfisterer19; clinic cohort
Pfisterer (≥75 years)0.25Betaα = 37; β = 109Pfisterer19; clinic cohort
Hospitalisation for heart failure risk ratio
Pfisterer (all ages)0.74lognormal95% CI = 0.48; 1.15Pfisterer19
Pfisterer <75 yrs0.53lognormal95% CI = 0.25; 1.15Pfisterer19
Pfisterer ≥75 yrs0.92lognormal95% CI = 0.57; 1.47Pfisterer19
Hospitalisation for heart failure; Baseline risk
Pfisterer all ages0.16betaα = 40; β = 208Pfisterer19; clinic cohort
Pfisterer <75 yrs0.16betaα = 16; β = 86Pfisterer19; clinic cohort
Pristerer ≥75 yrs0.20betaα = 29; β = 117Pfisterer19; clinic cohort
Drug usage
All ages
ACEI\ARB, baseline dose, Clinic0.50lognormalSE = 0.023Pfisterer19
ACEI\ARB, dose change, Clinic0.15lognormalSE = 0.063Pfisterer19
BB, dose change, Clinic0.14lognormalSE = 0.06Pfisterer19
ACEI\ARB, baseline dose, NP0.53lognormalSE = 0.026Pfisterer19
ACEI\ARB, dose change, NP0.27lognormalSE = 0.067Pfisterer19
BB, dose change, NP0.24lognormalSE = 0.054Pfisterer19
BB, baseline dose, Clinic0.25betaα = 62; β = 186Pfisterer19
BB, baseline dose, NP0.25betaα = 62.75; β = 188.25Pfisterer19
< 75 years
ACEI\ARB, dose change, Clinic0.16lognormalSE = 0.054Pfisterer19
BB, dose change, Clinic0.16lognormalSE = 0.06Pfisterer19
ACEI\ARB, dose change, NP0.29lognormalSE = 0.067Pfisterer19
BB, dose change, NP0.28lognormalSE = 0.054Pfisterer19
75 years
ACEI\ARB, dose change, Clinic0.15lognormalSE = 0.063Pfisterer19
BB, dose change, Clinic0.12lognormalSE = 0.041Pfisterer19
ACEI\ARB, dose change, NP0.25lognormalSE = 0.051Pfisterer19
BB, dose change, NP0.20lognormalSE = 0.052Pfisterer19
All patients/<75 years/75 years
Spironolactone, probability of use, Clinic0.23betaα = 56; β = 192Pfisterer19
Eplerenone, probability of use, Clinic0.40betaα = 100; β = 148Pfisterer19
Spironolactone, probability of use, NP0.30betaα = 76; β = 175Pfisterer19
Eplerenone, probability of use, NP0.41betaα = 103; β = 148Pfisterer19
Jourdain
Mortality risk ratio0.64lognormal95% CI = 0.26; 1.58Jourdain20
Mortality baseline risk0.10Betaα = 11; β = 99Jourdain20; clinic cohort
Hospitalisation for heart failure risk ratio0.46lognormal95% CI = 0.3; 0.7Jourdain20
Hospitalisation for heart failure; Baseline risk0.44betaα = 48; β = 62Jourdain20; clinic cohort
Treatement modification (biochemistry testing)
Diuretic, NP group0.5betaα = 55; β = 55Jourdain20
ACEI, NP group0.19betaα = 21; β = 89Jourdain20
Spironolactone\eplerenone, NP group0.54betaα = 196; β = 165Combined data from Jourdain20 and Pfisterer19
Diuretic, Clinic group0.24betaα = 26; β = 84Pfisterer19
ACEI, Clinic group0.08betaα = 9; β = 101Pfisterer19
Spironolactone\eplerenone, Clinic group0.46betaα = 163; β = 195Combined data from Jourdain20 and Pfisterer19
Drug usage
ACEI\ARB, baseline dose, Clinic18.8lognormalSE = 10 (assumed 50% of target dose as SE)Jourdain20
BB, baseline dose, Clinic28.5lognormalSE = 25 (assumed 50% of target dose as SE)Jourdain20
ACEI\ARB, 3 months dose, Clinic19.6lognormalSE = 10 (assumed 50% of target dose as SE)Jourdain20
BB, 3 months dose, Clinic33.5lognormalSE = 25 (assumed 50% of target dose as SE)Jourdain20
ACEI\ARB, baseline dose, NP18.8lognormalSE = 10 (assumed 50% of target dose as SE)Jourdain20
BB, baseline dose, NP29.0lognormalSE = 25 (assumed 50% of target dose as SE)Jourdain20
ACEI\ARB, 3 months dose, NP21.2lognormalSE = 10 (assumed 50% of target dose as SE)Jourdain20
BB, 3 months dose, NP38.5lognormalSE = 25 (assumed 50% of target dose as SE)Jourdain20
Furosemide, baseline dose, Clinic52.0lognormalSE = 5.72Jourdain20
Furosemide, dose change, Clinic9.0lognormalSE = 1.91Jourdain20
Furosemide, baseline dose, NP50.0lognormalSE = 4.58Jourdain20
Furosemide, dose change, NP9.0lognormalSE = 1.91Jourdain20
Troughton
Hospitalisation for heart failure (Risk ratio)0.42lognormal95% CI = 0.17; 1.05Troughton18
Outpatient visits
Additional outpatient visit per patient; Clinic group0.30lognormalSE = 0.15Troughton18
Additional outpatient visit per patient; NP group0.90lognormalSE = 0.45Troughton18
Drug usage
ACEI\ARB, baseline dose, Clinic13.1lognormalSE = 1.12Troughton18
Furosemide, baseline dose, Clinic87.0lognormalSE = 19.83Troughton18
ACEI\ARB, dose change, Clinic1.2lognormalSE = 1.15Troughton18
Furosemide, 6 months, Clinic141.0lognormalSE = 43.83Troughton18
ACEI\ARB, baseline dose, NP15.3lognormalSE = 1.38Troughton18
Furosemide, baseline dose, NP123.0lognormalSE = 25.24Troughton18
ACEI\ARB, dose change, NP4.8lognormalSE = 1.03Troughton18
Furosemide, 6 months, NP197.0lognormalSE = 41.26Troughton18
Spironolactone, probability of use, 6 months, Clinic0.028betaα = 1; β = 35Troughton18
BB, probability of use, baseline, Clinic0.028betaα = 1; β = 35Troughton18
BB, probability of use, 6 months, Clinic0.056betaα = 2; β = 34Troughton18
Spironolactone, probability of use, 6 months, NP0.18betaα = 6; β = 27Troughton18
BB, probability of use, baseline, NP0.12betaα = 4; β = 29Troughton18
BB, probability of use, 6 months, NP0.12betaα = 4; β = 29Troughton18
Patient with CHF and LVSD (meta-analysis of Pfisterer, Jourdain, and Troughton)
Mortality risk ratio
Meta-analysis of Pfisterer (all ages) and Jourdain0.70lognormal95% CI = 0.5; 0.99Pfisterer and Jourdain
Hospitalisation for heart failure risk ratio
Meta-analysis (Jourdain, Pfisterer, and Troughton)0.57lognormal95% CI = 0.42; 0.76Jourdain20, Pfisterer19, and Troughton18
Mean cost (£)
Hospitalisation cost
Elective Inpatient
Heart Failure or Shock with CC3954gammaSE = 2114 / α = 3.50; β = 1130.34 / Using interquartile range (20001; 4703)NHS reference cost30
Heart Failure or Shock without CC2756gammaSE = 1862 / α = 2.19; β = 1258.09 / Using interquartile range (1262; 3562)NHS reference cost30
Elective Inpatient Excess Bed Day
Heart Failure or Shock with CC186gammaSE = 56.5 / α = 10.79; β = 17.20 / Using interquartile range (113; 187)NHS reference cost30
Heart Failure or Shock without CC238gammaSE = 95.5 / α = 6.22; β = 38.29 / Using interquartile range (177; 302)NHS reference cost30
Non-Elective Inpatient (Long Stay) HRG Data
Heart Failure or Shock with CC2608gammaSE = 774 / α = 11.35; β = 229.73 / Using interquartile range (1949; 2976)NHS reference cost30
Heart Failure or Shock without CC1692gammaSE = 508 / α = 11.10; β = 152.50 / Using interquartile range (1268; 1942)NHS reference cost30
Non-Elective Inpatient (Long Stay) Excess Bed Days HRG Data
Heart Failure or Shock with CC193gammaSE = 59 / α = 10.67; β = 18.06 / Using interquartile range (152; 230)NHS reference cost30
Heart Failure or Shock without CC189gammaSE = 57 / α = 11.01; β = 17.18 / Using interquartile range (151; 228)NHS reference cost30
Non-Elective Inpatient (Short Stay) HRG Data
Heart Failure or Shock with CC356gammaSE = 120 / α = 8.81; β = 40.44 / Using interquartile range (248; 406)NHS reference cost30
Heart Failure or Shock without CC340gammaSE = 106 / α = 10.29; β = 33.04 / Using interquartile range (248; 388)NHS reference cost30
Cardiologist outpatient visit cost
Consultant Led: Follow up Attendance Non- Admitted Face to Face105gammaSE = 35.5 / α = 8.80; β = 11.97 / Using interquartile range (75; 122)NHS reference cost30
Consultant Led: Follow up Attendance Multiprofessional Non- Admitted Face to Face125gammaSE = 11 / α = 129.01; β = 0.97 / Using interquartile range (123; 138)NHS reference cost30
Non-Consultant Led: Follow up Attendance Non-Admitted Face to Face71gammaSE = 44 / α = 2.62; β = 27.19 / Using interquartile range (38; 93)NHS reference cost30
Non-Consultant Led: Follow up Attendance Multiprofessional Non- Admitted Face to Face117gammaSE = 27 / α = 18.85; β = 6.22 / Using interquartile range (85; 121)NHS reference cost30
Mean utility scores
NYHA class I0.855Beta95% CI = 0.845; 0.864 / α = 1391.94; β = 236.06Gohler 200928
NYHA class II0.771Beta95% CI = 0.761; 0.781 / α = 1255.19; β = 372.81Gohler 200928
NYHA class III0.673Beta95% CI = 0.665; 0.690 / α = 1095.64; β = 532.36Gohler 200928
NYHA class IV0.532Beta95% CI = 0.480; 0.584 / α = 866.1; β = 761.9Gohler 200928
Other
Standard Mortality ratios (Mean %)
Male, 65–74 years573Gamma95% CI = 521; 631 / SE = 30 / α = 364.81; β = 1.57Guili 200524
Male, 85+ years241Gamma213; 272 / SE = 15 / α = 258.14; β = 0.93Guili 200524
Female, 65–74 years718Gamma641; 804 / SE = 42 / α = 292.25; β = 2.46Guili 200524
Female, 85+ years242Gamma223; 262 / SE = 14 / α = 298.80; β = 0.81Guili 200524
Effect of ACEI on survival (Risk ratio)0.86lognormal95% CI = 0.81; 0.91Flather 200025
Effect of BB on survival (Risk ratio)0.57lognormal95% CI = 0.51; 0.64Shibata 200126
Biochemistry test cost (£)1.34gammaSE = 0.59 / α = 5.16; β = 0.26 / Using interquartile range (0.79; 1.56)NHS reference cost30

NP = Natriuretic Peptide; Clinic = Clinical assessment; UC = Usual Care

12. References

1.
Kinnunen P, Vuolteenaho O, Ruskoaho H. Mechanisms of atrial and brain natriuretic peptide release from rat ventricular myocardium: effect of stretching. Endocrinology. 1993;132(5):1961–1970. [PubMed: 8477647]
2.
Yasue H, Yoshimura M, Sumida H, et al. Localization and mechanism of secretion of B-type natriuretic peptide in comparison with those of A-type natriuretic peptide in normal subjects and patients with heart failure. Circulation. 1994;90(1):195–203. [PubMed: 8025996]
3.
Hunt PJ, Espiner EA, Nicholls MG, et al. The role of the circulation in processing pro-brain natriuretic peptide (proBNP) to amino-terminal BNP and BNP-32. Peptides. 1997;18(10):1475–1481. [PubMed: 9437705]
4.
Richards AM, Nicholls MG, Yandle TG, et al. Plasma N-terminal pro-brain natriuretic peptide and adrenomedullin: new neurohormonal predictors of left ventricular function and prognosis after myocardial infarction. Circulation. 1998;97(19):1921–1929. [PubMed: 9609085]
5.
Richards AM, Nicholls MG, Yandle TG, et al. Neuroendocrine prediction of left ventricular function and heart failure after acute myocardial infarction. The Christchurch Cardioendocrine Research Group. Heart. 1999;81(2):114–120. [PMC free article: PMC1728940] [PubMed: 9922344]
6.
Maisel AS, Krishnaswamy P, Nowak RM, et al. Rapid measurement of B-type natriuretic peptide in the emergency diagnosis of heart failure. New England Journal of Medicine. 2002;347(3):161–167. [PubMed: 12124404]
7.
McCullough PA, Nowak RM, McCord J, et al. B-type natriuretic peptide and clinical judgment in emergency diagnosis of heart failure: analysis from Breathing Not Properly (BNP) Multinational Study. Circulation. 2002;106(4):416–422. [PubMed: 12135939]
8.
Stanek B, Frey B, Hulsmann M, et al. Prognostic evaluation of neurohumoral plasma levels before and during beta-blocker therapy in advanced left ventricular dysfunction. Journal of the American College of Cardiology. 2001;38(2):436–442. [PubMed: 11499735]
9.
Koglin J, Pehlivanli S, Schwaiblmair M, et al. Role of brain natriuretic peptide in risk stratification of patients with congestive heart failure. Journal of the American College of Cardiology. 2001;38(7):1934–1941. [PubMed: 11738297]
10.
Bettencourt P, Ferreira A, Dias P, et al. Predictors of prognosis in patients with stable mild to moderate heart failure. J Card Fail. 2000;6(4):306–313. [PubMed: 11145755]
11.
Gardner RS, Ozalp F, Murday AJ, et al. N-terminal pro-brain natriuretic peptide. A new gold standard in predicting mortality in patients with advanced heart failure. Eur Heart J. 2003;24(19):1735–1743. [PubMed: 14522568]
12.
Yoshimura M, Mizuno Y, Nakayama M, et al. B-type natriuretic peptide as a marker of the effects of enalapril in patients with heart failure. American Journal of Medicine. 2002;112(9):716–720. [PubMed: 12079712]
13.
Ferreira A, Bettencourt P, Dias P, et al. Neurohormonal activation, the renal dopaminergic system and sodium handling in patients with severe heart failure under vasodilator therapy. Clinical Science (London) 2001;100(5):557–566. [PubMed: 11294697]
14.
Murdoch DR, McDonagh TA, Byrne J, et al. Titration of vasodilator therapy in chronic heart failure according to plasma brain natriuretic peptide concentration: randomized comparison of the hemodynamic and neuroendocrine effects of tailored versus empirical therapy. American Heart Journal. 1999;138(6 Pt 1):1126–1132. [PubMed: 10577444]
15.
Latini R, Masson S, Anand I, et al. Effects of valsartan on circulating brain natriuretic peptide and norepinephrine in symptomatic chronic heart failure: the Valsartan Heart Failure Trial (Val-HeFT) Circulation. 2002;106(19):2454–2458. [PubMed: 12417542]
16.
Tsutamoto T, Wada A, Maeda K, et al. Effect of spironolactone on plasma brain natriuretic peptide and left ventricular remodeling in patients with congestive heart failure. Journal of the American College of Cardiology. 2001;37(5):1228–1233. [PubMed: 11300427]
17.
Fung JW, Yu CM, Yip G, et al. Effect of beta blockade (carvedilol or metoprolol) on activation of the renin-angiotensin-aldosterone system and natriuretic peptides in chronic heart failure. American Journal of Cardiology. 2003;92(4):406–410. [PubMed: 12914870]
18.
Troughton RW, Frampton CM, Yandle TG, et al. Treatment of heart failure guided by plasma aminoterminal brain natriuretic peptide (N-BNP) concentrations. Lancet. 2000;355(9210):1126–1130. [PubMed: 10791374]
19.
Pfisterer M, Buser P, Rickli H, et al. BNP-guided vs symptom-guided heart failure therapy: the Trial of Intensified vs Standard Medical Therapy in Elderly Patients With Congestive Heart Failure (TIME-CHF) randomized trial. Journal of the American Medical Association. 2009;301(4):383–392. [PubMed: 19176440]
20.
Jourdain P, Jondeau G, Funck F, et al. Plasma brain natriuretic peptide-guided therapy to improve outcome in heart failure: the STARS-BNP Multicenter Study. Journal of the American College of Cardiology. 2007;49(16):1733–1739. [PubMed: 17448376]
21.
Lainchbury JG, Troughton RW, Strangman KM, et al. NTproBNP guided treatment for chronic heart failure: results from the Battlescarred trial. Journal of the American College of Cardiology. 2010;55(1):53–60. [PubMed: 20117364]
22.
Morimoto T, Hayashino Y, Shimbo T, et al. Is B-type natriuretic peptide-guided heart failure management cost-effective? International Journal of Cardiology. 2004;96(2):177–181. [PubMed: 15262031]
23.
Beck-da-Silva L, de BA, Fraser M, et al. BNP-guided therapy not better than expert’s clinical assessment for beta-blocker titration in patients with heart failure. Congestive Heart Failure. 2005;11(5):248–253. [PubMed: 16230866]
24.
de Guili F, Khaw K-T, Cowie MR, et al. Incidence and outcome of persons with a clinical diagnosis of heart failure in a general practice population of 696,884 in the United Kingdom. European Journal of Heart Failure. 2005;7:295–302. [PubMed: 15718168]
25.
Flather MD, Yusuf S, Kober L, et al. Long-term ACE-inhibitor therapy in patients with heart failure or left-ventricular dysfunction: a systematic overview of data from individual patients. ACE-Inhibitor Myocardial Infarction Collaborative Group. Lancet. 2000;355(9215):1575–1581. [PubMed: 10821360]
26.
Shibata MC, Flather MD, Wang D. Systematic review of the impact of beta blockers on mortality and hospital admissions in heart failure. European Journal of Heart Failure. 2001;3(3):351–357. [PubMed: 11378007]
27.
Government Actuary’s Department. Interim Life tables. [Last accessed on: 2009 Nov. 23]. Available from: Government Actuary’s Department.
28.
Gohler A, Geisler BP, Manne JM, et al. Utility estimates for decision-analytic modeling in chronic heart failure - Health states based on New York Heart Association classes and number of rehospitalizations. Value in Health. 2009;12(1):185–187. [PubMed: 18647251]
29.
Pitt B, Remme W, Zannad F, et al. Eplerenone, a selective aldosterone blocker, in patients with left ventricular dysfunction after myocardial infarction. New England Journal of Medicine. 2003;348(14):1309–1321. [PubMed: 12668699]
30.
Department of Health. NHS reference costs 2007–08. UK Department of Health; 2009. http://www​.dh.gov.uk​/en/Publicationsandstatistics​/Publications​/PublicationsPolicyAndGuidance​/DH_098945.
31.
British Medical Association and Royal Pharmaceutical Society of Great Britain. British National Formulary. 58. London:UK: BMJ Group and RPS Publishing; 2009.
32.
Swedberg K, Cleland J, Dargie H, et al. Guidelines for the diagnosis and treatment of chronic heart failure: executive summary (update 2005): The Task Force for the Diagnosis and Treatment of Chronic Heart Failure of the European Society of Cardiology. Eur Heart J. 2005;26(11):1115–1140. [PubMed: 15901669]
33.
Hunt SA, Abraham WT, Chin MH, et al. ACC/AHA 2005 Guideline Update for the Diagnosis and Management of Chronic Heart Failure in the Adult: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure): Developed in Collaboration With the American College of Chest Physicians and the International Society for Heart and Lung Transplantation: Endorsed by the Heart Rhythm Society. Circulation. 2005;112(12):e154–e235. [PubMed: 16160202]
34.
Personal Social Services Research Unit. Unit Costs of Health and Social Care 2008. Canterbury: UK: Personal Social Services Research Unit; 2008.
35.
Stewart S, Jenkins A, Buchan S, et al. The current cost of heart failure to the National Health Service in the UK. European Journal of Heart Failure. 2002;4(3):361–371. [PubMed: 12034163]

Footnotes

b

Beck-da-Silva published in 2005 results from a RCT23 assessing serial measurement of natriuretic peptide concentration for beta-blocker up-titration as opposed to monitoring the entire drug usage in Troughton 200018, Jourdain 200720, Pfisterer 200919, and Lainchbury 201021. For this reason, and considering that Beck-da-Silva trial23 has a small cohort size (N=41) and did not reported sensible outcomes for economic modelling (all-cause mortality and heart failure-related hospitalizations), this study was not utilized for this economic analysis.

c

The Pfisterer trial19 follow-up (18 months) was the longest of the meta-analysed trials (15 months for Jourdain20 and 9.5 months for Troughton18). Troughton21 did not report all-cause mortality but only cardiovascular deaths.

d

45% men (n=2884), mean age 75 years (SD=9); 55% women (n=3594), mean age 79 years (SD=9).

e

Effect of ACEI versus placebo: RR=0.86 (95% CI 0.81–0.91); Effect of BB versus placebo: RR=0.57 (95% CI 0.51–0.64); Combined effect of ACEI and BB = 0.4902.

f

Pfisterer assessed patients’ quality of life using the Minnesota Living with Heart Failure questionnaire, the Duke Activity Status Index, and the Short Form 12, reporting no significant differences in the magnitude of improvements between strategies. Lainchbury administered the Minnesota Living with Heart Failure questionnaire and showed that Minnesota scores improved significantly and similarly in natriuretic peptide and clinical assessment cohorts. Troughton reported that quality of life scores remained stable for compared cohorts of patients.

g

Study selected from a non-systematic search for utility scores in CHF. The Gohler paper was selected as being a recent assessment estimating utility scores in CHF using EQ-5D data collected from a well-recognized RCT on patients with CHF.

h

The Pfisterer trial19 follow-up (18 months) was the longest of the meta-analysed trials (15 months for Jourdain20 and 9.5 months for Troughton18).

i

A weighted average cost was calculated considering elective and non-elective inpatient admissions for heart failure. Excess bed days were added to this calculation30.

j

The drug usage was reported for BB in metoprolol equivalent. Metoprolol is an available treatment in the UK, but not licensed for use in heart failure31. We costed metoprolol to be consistent with clinical trial outcomes. We consider this is not likely to affect the applicability of our results in a UK context.

k

Target doses as recommended by the European Society of Cardiology32, 33, referred to in the RCTs19, 20.

l

The baseline usage was the same for compared cohorts

m

In sensitivity analyses, Troughton18 and Jourdain20 drug usages were applied to the cost-effectiveness assessment developed on patients with CHF and LVSD based on Pfisterer19, Jourdain20, and Troughton18. For this assessment, outcomes from trials were assumed at 18 months, the Pfisterer follow-up19 being the longest one (15 months for Jourdain20 and 9.5 months for Troughton18).

n

A weighted average cost was calculated considering cardiology follow-up visits (not leading to admission), by consultant and non-consultant, with or without a multiprofessional approach.

o

Test costs are equivalent for BNP and NT-proBNP

p

For cost-effectiveness assessments based on Lainchbury21, four natriuretic peptide tests were assumed during year two and were discounted (as for outpatient visits – Section 5.3)

q

£905 million (1.91% of total NHS expenditure); 1.01 million cases35.

r

(1) Hospitalisation data: (a) Pfisterer19 (all ages) baseline risk (clinical assessment cohort) = 0.16; RR (natriuretic peptide vs clinical assessment) = 0.74 [0.48; 1.15]. (b) Jourdain20 baseline risk = 0.44; RR = 0.46 [0.3; 0.7]. (2) Mortality: (a) Pfisterer19 baseline risk = 0.22; RR = 0.72 [0.5; 1.04]. (b) Jourdain20 baseline risk = 0.10; RR = 0.64 [0.26; 1.58]. (c) We used area under curves for Pfisterer19 main analysis (all ages) to estimate life years instead of end-of-trial RR as in the combined analysis (CHF and LVSDSection 4.1.1).

s

Troughton18 did not report all-cause mortality; has a small cohort size (N=69); and this trial was conducted before BB were commonly used in CHF. We considered that modelling Troughton18 independently would not add value to this economic analysis.

t

We assumed the same mortality rate and yearly cost per patient up to three years after trial periods.

u

Due to a ‘bug’, excel cannot calculate the gamma distribution when the standard error is very small compared with the mean. This was the case with some mean drug dosage and therefore we used the lognormal distribution instead.

Copyright © 2010, National Clinical Guideline Centre.

Apart from any fair dealing for the purposes of research or private study, criticism or review, as permitted under the Copyright, Designs and Patents Act, 1988, no part of this publication may be reproduced, stored or transmitted in any form or by any means, without the prior written permission of the publisher or, in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK. Enquiries concerning reproduction outside the terms stated here should be sent to the publisher at the UK address printed on this page.

Bookshelf ID: NBK65326

Views

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...