Aims and objectives
Our aim was to assess the net health effects and cost-effectiveness of lifetime statin therapy of different intensity in the contemporary UK population from the perspective of the UK NHS.
To achieve this, we used the CVD microsimulation policy model with integrated healthcare costs and QoL models to project CVD disease progression, QoL-adjusted life expectancy, and healthcare costs with and without statin treatment.
Methods
The cardiovascular disease microsimulation policy model
The CVD microsimulation policy model (see Chapter 3) with integrated primary care and hospital inpatient care cost models (see Chapter 4) and QoL model (see Chapter 5) informed the assessment of the cost-effectiveness of statin therapy. The model was used to project event risks, survival, primary and hospital care costs and QoL and to summarise life years, QALYs and healthcare costs over individuals’ remaining lifetimes (i.e. until death or 110 years of age) without and with statin treatment, and to assess the cost-effectiveness of statin therapies in categories of individuals.
Health-related quality of life
We estimated the HRQoL associated with participant characteristics, disease histories and events by employing a linear regression model using data from the HSE conducted in 2006, 2011 and 2017 (see Chapter 5), which measured participants’ QoL using the EQ-5D questionnaire. We valued participants’ QoL using UK valuations of QoL utility.46,47 The estimated QoL model was then integrated into the CVD model to forecast individuals’ QoL in each year in the model.
Effects and costs of statin therapy
Table 13 presents the effects and costs of statin therapy used in the analyses. We evaluated the effects of standard statin therapy (achieving 35–45% reduction in LDL-C) and higher-intensity statin therapy (achieving ≥ 45% LDL-C reduction) (see Table 1) compared with no statin treatment. The absolute reduction in LDL-C was calculated based on the proportional reduction achieved by the statin regimen and the individual’s pre-treatment LDL-C level. Reductions in cardiovascular event risks with statin regimens were calculated based on the absolute reduction in LDL-C with the particular statin regimen and the relative reduction of cardiovascular event risks per 1 mmol/l LDL-C reduction with statin therapy, reported by the CTTC IPD meta-analysis of randomised trials of statin therapy.5 The analyses took into account the excess rates of new-onset diabetes,12,13 myopathy and rhabdomyolysis60 reported with statin therapy (see Table 13).
| Item | Value | Source |
|---|---|---|
| Effects of statin therapy on cardiovascular events per 1-mmol/l reduction in LDL-C, RR (95% CI) | CTTC IPD meta-analysis5 | |
| MI | 0.76 (0.73 to 0.79) | |
| Stroke | 0.84 (0.80 to 0.89) | |
| CRV | 0.75 (0.73 to 0.78) | |
| VD | 0.88 (0.85 to 0.91) | |
| Adverse effects of statin therapy on: | ||
| Incident diabetes, OR (95% CI) | ||
| With standard statin therapy compared with no statin treatment | 1.09 (1.02 to 1.17) | Meta-analyses of randomised controlled trials12 |
| With higher-intensity statin therapy compared with standard statin therapy | 1.12 (1.04 to 1.22) | Meta-analyses of randomised controlled trials13 |
| Myopathy | ||
| Excess per 100,000 person-years on statin therapy (95% CI) | 11 (4 to 27) | Overview of cohort studies60 |
| Occurrence of myopathy is associated with reduction in QoL over 30 days’ recovery period. Statin treatment is stopped | 0.017 QALY reduction in year | Modelling study61 |
| Rhabdomyolysis | ||
| Excess per 100,000 person-years on statin therapy (95% CI) | 3.4 (1.6 to 6.5) | Overview of cohort studies60 |
| Case fatality | 10% | Overview of cohort studies60 |
| Reduction in QoL | 50% over 7.5 days hospital admission and by 20% for further 30 days recovery | Modelling study61 |
| LDL-C reductions with statin therapy: | ||
| With standard statin therapy (e.g. atorvastatin 20 mg/day, rosuvastatin 5–10 mg/day or simvastatin 40–80 mg/day) | 37–43%; 43% used in base-case | Meta-analysis of randomised controlled trials27 |
| With higher-intensity statin therapy (e.g. atorvastatin 40–80 mg/day, rosuvastatin 20–80 mg/day) | 48–58%; 55% used in base-case | Meta-analysis of randomised controlled trials27 |
| Statin therapy costs (£) | ||
| Standard statin therapy (e.g. atorvastatin 20 mg/day, rosuvastatin 5-10 mg/day or simvastatin 40–80 mg/day) | £14.09–19.57 per year; £14.35 used in base-case | NHS drug tariff, December 202114 |
| Higher-intensity statin therapy (e.g. atorvastatin 40–80 mg/day, rosuvastatin 20–40 mg/day) | £15.91–27.91 per year; £21.91 used in base-case | NHS drug tariff, December 202114 |
| Statin initiation and monitoring healthcare costs (£) | ||
| • In year of initiation (a doctor and a nurse consultations; tests of blood lipids, HbA1c, thyroid function) | £54.65 | Unit Costs of Health and Social Care;37 NHS reference costs62 |
| • In subsequent years: a nurse consultation and a blood lipids test (for people with previous CVD) | £12.05 | Unit Costs of Health and Social Care;37 NHS reference costs62 |
Cost of statin treatment included the cost of generic statin medication14 and the costs of consultations37 and blood lipid tests62 for the initiation and monitoring of statin prescribing in the NHS (see Table 13).
Study population
UK Biobank individual participants’ characteristics at entry into the study informed the analyses of statin cost-effectiveness in people 40–70 years old. We report results in categories of participants by history of CVD, sex, age, untreated LDL-C level, and, for those without previous CVD at entry, by their estimated 10-year CVD risk, calculated using the QRISK3 risk score.21
Cost-effectiveness analyses
Base-case analysis
The cost-effectiveness of lifetime statin therapy was evaluated from the perspective of the UK NHS. The following assumptions were made: (1) the reduction in LDL-C levels with specific statin regimen corresponded to the average proportional reduction achieved with that regimen and the pre-treatment LDL-C level, (2) the relative effects of statin therapy on event risks were similar in categories of participants and remained unchanged over the duration of therapy, (3) disease events did not differ in severity irrespective of treatment regimen and (4) statin therapy did not affect the risks of cancer or other nonvascular events63 (except the risk of incident diabetes, myopathy and rhabdomyolysis as the adverse events), nor cause any discomfort or disutility beyond the CVD events and adverse events previously specified.
For each individual, we ran 500 microsimulations to address the first-order uncertainty. We reported the gained life-years and QALYs, additional statin and healthcare costs, and incremental costs per QALY for standard and higher-intensity statin therapy. The analysis followed the NICE manual for health technology evaluations with future life-years, QALYs and costs discounted at 3.5% per year.64
Assessment of parameter uncertainty
We summarised parameter uncertainty using 500 and 1000 sets of parameter values, including uncertainty in statin effects on event risks, risk equations, QoL and healthcare costs, for participants without and participants with previous CVD history, respectively. Values for treatment effects were sampled from log-normal distributions based on the relative risk reductions with statin therapy (see Table 13). Bootstrapping was used to derive the values for (1) the parameters of the events risk equations and of the healthcare cost equations from the UKB study population and (2) the parameters of the QoL equation from the HSE participants’ data. Cost-effectiveness acceptability curves and probability of cost-effectiveness at willingness-to-pay thresholds of £20,000 and £10,000 per QALY were reported.
Sensitivity and scenario analyses
We varied key parameters in sensitivity analyses. First, we assumed the relative risk reductions of cardiovascular events with statin therapy to increase annually, as suggested by Mendelian randomisation studies,65 with a further 1.5% proportional reduction per 1 mmol/l reduction in LDL-C each year following the first 5 years of treatment. In another scenario analysis, we assumed the base-case relative risk reduction to decline by 10% annually from year 6 onwards. Second, we applied smaller relative risk reductions in cardiovascular events per 1 mmol/l LDL-C after 75 years of age, informed from data only among people aged > 75 years in the IPD meta-analysis.5 Third, we ran scenario analyses with small detrimental or beneficial statin effect on incident cancer, informed by the 95% CI limits reported in an IPD meta-analysis of randomised statin trials.63 Fourth, we implemented a scenario analysis that assessed statin cost-effectiveness under real-world compliance with statin therapy over time, derived from routine UK data,66 with statin effects and costs discontinued with therapy discontinuation. Fifth, we included analyses with hypothetical extra disutility associated with statin treatment equal to 0.001 or 0.002 each year.67 In further sensitivity analyses, we reduced the effects of cardiovascular events or diabetes adverse effects on QoL by 50% and used discount rates of 1.5% instead of 3.5% per year. We also presented results of a scenario analysis in which only healthcare costs for CVD and incident diabetes were included and a sensitivity analysis with higher costs of statin therapy.
Further details of the methods of cost-effectiveness analyses are available in Appendix 6, including further details of the sensitivity and scenario analyses (see Appendix 6, Tables 30 and 31).
Results
Categories of people 40–70 years old
Table 14 presents the number of UKB participants in categories. Participants who were identified to have a lower risk of CVD were predominantly women, younger, and with lower LDL-C levels. There were no men aged 60–70 years with a 10-year CVD risk of ˂ 5%. A small number of men and women aged 40–49 years with a 10-year CVD risk ˃ 15% were grouped together with those with a 10-year CVD risk of 10–15% in the corresponding categories by LDL-C for the presentation of results.
| Sex, age (years) | Without CVD, by 10-year CVD risk (QRISK3, %) | With CVD | ||||
|---|---|---|---|---|---|---|
| < 5 | 5–10 | 10–15 | 15–20 | ≥ 20 | ||
| Pre-treatment LDL-C (mmol/l): < 3.4 | ||||||
| Men | ||||||
| 40–49 | 13,497 | 4037 | 711 | 239a | 228a | 939 |
| 50–59 | 2281 | 9055 | 4759 | 1847 | 1622 | 2263 |
| 60–70 | 0 | 1535 | 6073 | 6513 | 9113 | 6218 |
| Women | ||||||
| 40–49 | 32,831 | 986 | 146a | 60a | 119a | 1430 |
| 50–59 | 18,843 | 6707 | 1025 | 308a | 366a | 1868 |
| 60–70 | 1522 | 9822 | 5742 | 1994 | 1589 | 2956 |
| Pre-treatment LDL-C (mmol/l): ≥ 3.4, < 4.1 | ||||||
| Men | ||||||
| 40–49 | 10,436 | 5629 | 1099 | 332a | 289a | 853 |
| 50–59 | 1255 | 9445 | 7190 | 3079 | 2408 | 2317 |
| 60–70 | 0 | 908 | 5919 | 8003 | 14,028 | 6607 |
| Women | ||||||
| 40–49 | 17,285 | 1214 | 178a | 59a | 86a | 827 |
| 50–59 | 17,471 | 11,228 | 1980 | 540 | 454a | 2109 |
| 60–70 | 1304 | 14,829 | 11,580 | 4693 | 3039 | 4298 |
| Pre-treatment LDL-C (mmol/l): ≥ 4.1 | ||||||
| Men | ||||||
| 40–49 | 5623 | 5310 | 1605 | 645 | 660 | 1006 |
| 50–59 | 423a | 5790 | 6616 | 4044 | 5005 | 3581 |
| 60–70 | 0 | 260 | 2891 | 5634 | 18,960 | 9950 |
| Women | ||||||
| 40–49 | 6943 | 1247 | 295a | 128a | 214a | 557 |
| 50–59 | 10,022 | 12,220 | 3574 | 1280 | 1300 | 2645 |
| 60–70 | 568 | 12,079 | 14,068 | 8507 | 9135 | 6854 |
Base-case cost-effectiveness of statin therapy in categories of 40–70 years old
In participant categories defined by age, sex, previous CVD and 10-year CVD risk, standard statin therapy (20 mg of atorvastatin daily) was projected to increase individual QALYs (undiscounted) by 0.20–1.09, and higher-intensity statin therapy (80 mg of atorvastatin daily) by a further 0.03–0.20 QALYs (see Figure 8). Across population categories, standard statin therapy compared with no statin therapy had an incremental cost per QALY ranging from £280 to £8530, with higher-intensity statin therapy realising additional QALYs compared with standard statin therapy at an incremental cost per QALY ranging from £2610 to £47,640 (see Figure 9 and Appendix 7, Table 32). At a £20,000-per-QALY threshold, it was certain that either standard or higher-intensity statin therapy was cost-effective (see Appendix 7, Table 32). Higher-intensity statin therapy had the higher probability of being cost-effective in most participant categories and standard statin therapy was most likely to be cost-effective in categories of younger participants with lower LDL-C and/or lower CVD risk levels (see Figure 10). The probability that either standard or higher-intensity statin therapy was cost-effective remained above 80% at the £10,000-per-QALY threshold across all participant categories, with higher-intensity statin therapy having higher probability of being cost-effective at higher CVD risk and/or higher LDL-C levels (see Figure 10 and Appendix 7, Table 32).
FIGURE 8 Undiscounted QALYs gained with lifetime statin therapy in categories by sex, age, CVD risk and pre-treatment LDL-C level. NA, not applicable (no UKB participants in this category). (Figure not reproduced here; see the original article.)
FIGURE 9 Incremental cost per QALY gained (£/QALY gained) with lifetime statin therapy in categories by sex, age, CVD risk and pre-treatment LDL-C level. ICER, incremental cost-effectiveness ratio with costs and QALYs discounted at 3.5% per year; NA, not applicable (no UKB participants in this category). (Figure not reproduced here; see the original article.)
FIGURE 10 Probability of cost-effectiveness of lifetime statin therapy in categories by sex, age, CVD risk and pre-treatment LDL-C level. (a) At £20,000-per-QALY-gained threshold for cost-effectiveness and (b) at £10,000-per-QALY-gained threshold for cost-effectiveness. NA, not applicable (no UKB participants in this category) (Figure not reproduced here; see the original article.)
Sensitivity analyses of statin cost-effectiveness in categories of 40–70 years old
Sensitivity analyses found robust cost-effectiveness results across most categories of participants, except for the participants who were younger and had low 10-year CVD risk (Table 15). For these participants, the cost-effectiveness results were moderately impacted by the following hypothetical assumptions: (1) declining treatment effect after 5 years on statin treatment; (2) hypothetical increase in cancer incidence risk with statin treatment; (3) increased cost of statin medication and (4) extra QoL disutility with statin treatment ≥ 0.002 per year.
| People without CVD | ||||||||
|---|---|---|---|---|---|---|---|---|
| Sex, CVD risk | Standard statin vs. no statin therapy | Higher-intensity vs. standard statin therapy | ||||||
| Men, high | Men, moderate | Women, high | Women, moderate | Men, high | Men, moderate | Women, high | Women, moderate | |
| Age (years) | 60–70 | 40–49 | 60–70 | 40–49 | 60–70 | 40–49 | 60–70 | 40–49 |
| 10-year CVD risk (%) | 15–20 | 5–10 | 15–20 | 5–10 | 15–20 | 5–10 | 15–20 | 5–10 |
| LDL-C (mmol/l) | ≥ 4.1 | < 3.4 | ≥ 4.1 | < 3.4 | ≥ 4.1 | < 3.4 | ≥ 4.1 | < 3.4 |
| Base-case (£/QALY gained) | 840 | 3420 | 1480 | 6560 | 3620 | 18,970 | 5800 | 47,640 |
| Relative reduction in CVD events with statin therapy increasing annually (£) | 1300 | 2360 | 1640 | 4150 | 2750 | 7800 | 4320 | 12,370 |
| Relative reduction in CVD events with statin therapy decreasing annually (£) | 1290 | 20,040 | 2800 | 42,850 | 12,480 | –23,820a | 24,940 | –22,690a |
| Relative reduction in CVD events with statin therapy reduced in elderly (£) | 1120 | 3790 | 1990 | 7410 | 5170 | 28,220 | 8370 | 135,170 |
| RR in incident cancer with statin therapy of 0.96 (£) | 390 | 2120 | 1090 | 4230 | 3590 | 18,360 | 5750 | 47,000 |
| RR in incident cancer with statin therapy of 1.05 (£) | 1790 | 9720 | 2230 | 18,340 | 3660 | 19,680 | 5850 | 49,190 |
| Risk of NVD increased by 20% (£) | 650 | 3420 | 1400 | 6700 | 3550 | 20,560 | 5840 | 57,920 |
| Compliance with statin therapy as in routine care (£) | 1260 | 4520 | 1920 | 7480 | 4800 | 33,030 | 7500 | 101,540 |
| With QoL disutility of daily statin pill of 0.001/year (£) | 910 | 4390 | 1610 | 9440 | 3630 | 19,010 | 5810 | 47,830 |
| With QoL disutility of daily statin pill of 0.002/year (£) | 980 | 6130 | 1770 | 16,790 | 3630 | 19,060 | 5820 | 48,020 |
| QoL disutilities of CVD events reduced by 50% (£) | 870 | 3550 | 1540 | 6990 | 3790 | 20,690 | 6150 | 59,350 |
| QoL disutilities of CVD events increased by 50% (£) | 820 | 3300 | 1430 | 6190 | 3470 | 17,520 | 5490 | 39,790 |
| QoL disutilities of diabetes reduced by 50% (£) | 830 | 3280 | 1450 | 6130 | 3360 | 13,760 | 5160 | 25,390 |
| Discount rates for costs and outcomes at 1.5% per annum (£) | 1420 | 3140 | 1910 | 5700 | 3710 | 13,500 | 5420 | 28,210 |
| Include only healthcare costs for CVD and incident diabetes (£) | (–400) | 2120 | 190 | 4490 | 2260 | 17,040 | 4280 | 42,940 |
| Cost of statin × 1.5 (£) | 1360 | 5010 | 2060 | 8750 | 5300 | 26,240 | 7680 | 63,090 |
| Cost of statin × 2 (£) | 1890 | 6590 | 2640 | 10,930 | 6980 | 33,510 | 9570 | 78,530 |
| Cost of statin × 5 (£) | 5020 | 16,130 | 6120 | 24,040 | 17,050 | 77,140 | 20,900 | 171,210 |
Summary of findings
This study evaluated the cost-effectiveness of statin therapy in categories of men and women aged 40–70 years in the UK, based on their CVD risk and pre-treatment LDL-C levels. We report that lifetime standard statin therapy increased QoL adjusted survival in all categories studied and, at current UK cost of generic statin therapy, was highly cost-effective. Higher-intensity statin therapy was also found to be cost-effective in many categories with higher CVD risk or higher pre-treatment LDL-C levels. Sensitivity analyses confirmed the robustness of the results. The study highlights the need to improve statin uptake among eligible people and consider widening statin eligibility to optimise the benefits from statin therapy in the population.
Report Supplementary Material 1 includes the Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) Checklist.