Number of patients needed to prescribe statins in primary cardiovascular prevention: mirage and reality

Number of patients needed to prescribe statins in primary cardiovascular prevention: mirage and... Abstract Background Number of patients needed to treat (NNT) with a statin in primary prevention of coronary heart disease (CHD) is often misinterpreted because this single statistic averages results from heterogeneous studies. Objective To provide estimates of the number of individuals needed to be prescribed a statin to prevent one CHD event accounting for their level of CHD risk and for persistence to treatment. Methods A post hoc analysis was conducted based on a Cochrane review on statins for the primary prevention of cardiovascular diseases. Five-year NNTs were calculated separately from randomized clinical trials (RCTs), including ‘lower’ and ‘higher’ risk populations (CHD mean event rates of 3.7 and 14.4 per 1000 person–years, respectively). NNTs were adjusted for 5-year persistence to treatment using a value of 65%. Results Persistence-adjusted 5-year NNTs to prevent one CHD for the lower and higher CHD risk categories were 146 [95% confidence interval (CI): 117–211] and 53 (95% CI: 39–88) respectively, values 25% and 15% higher than their unadjusted counterpart (117, 95% CI: 94–167 and 46, 95% CI: 34–78). Conclusions Five-year NNTs for statins to prevent a first CHD is almost three times higher in those at lower versus higher risk populations. Reporting combined results from RCTs including subjects at different cardiovascular risks should be avoided. Individualizing the risk of CHD should orient family physicians and their patients in their choice of preventive approaches and generate more realistic expectations about compliance and outcomes. Adherence persistence, coronary artery disease, statins primary prevention, number needed to treat Introduction Number needed to treat (NNT) is often used in medical publications to translate complex research results into a simple metric easy to interpret and use in medical decision making. For instance, a 5-year NNT of 88 is interpreted as the estimated number of patients needed to be treated with statins over 5 years to prevent the occurrence of a first coronary heart disease (CHD) event (1). Although generally acknowledged, the actual impact of lower persistence to statin treatment in real-life compared with research populations from which NNT values are derived has not been adequately accounted for in NNT estimations (2–3). As the effectiveness of interventions to improve persistence to statins has been shown to be modest, it is imperative that clinicians, patients and decision makers have access to the right information, so they can base their treatment and health policy decisions on realistic expectations (4–5). We hypothesize that an NNT of 88 derived from meta-analysis of randomized clinical trials (RCTs) of statins for primary cardiovascular (CV) prevention does not reflect the true number of individuals needed to be prescribed a statin to prevent a first CHD event, because it includes populations at different CV risks with diverse outcomes, and because the drop-out rate from treatment is higher in real-life than in RCTs (6–7). The objective of this study was to provide estimates of the 5-year number of individuals needed to be prescribed a statin to prevent a first non-fatal or fatal myocardial infarction (MI) or CHD death accounting for their level of CHD risk and for real-life persistence to statin treatment in primary care. Methods This post hoc analysis used the results of nine RCTs included in a Cochrane meta-analysis of statins for the primary prevention of CV disease, that specifically reported outcomes on fatal and non-fatal MI and CHD deaths (supplementary Table) (6). CHD outcomes based on symptoms (angina and unstable angina) and coronary revascularization were excluded from this analysis as they may not have been uniformly defined and systematically reported between RCTs (i.e. ‘soft’ outcomes) (8). The nine RCTs were classified in two broad CHD-risk categories labelled ‘lower’ and ‘higher’, primarily based on the ranking of CHD event rates observed in the control group of each RCT. The ranking showed a gap between the ACAPS study with an event rate of 6.9 per 1000 person–years (p–y) and the KAPS study with a rate of 12.6 per 1000 p–y, which was used to delineate our two categories (Table 1). The PHYLLIS study (event rate of 4.6 per 1000 p–y) was reclassified in the higher CHD risk category because it included only patients with uncontrolled or untreated hypertension. The number of CHD events (fatal and non-fatal MI, and CHD deaths) in statin and control groups for each RCT were entered in a meta-analysis table stratified by the two CHD risk categories using Review Manager (RevMan) version 5.3 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). Table 1. Characteristics of RCTs included in the analysis, NNT and persistence-adjusted NNT by CHD-risk categoriesa CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD, coronary heart disease; CVD, cardiovascular diseases; F, female; HBP, high blood pressure and SBP, systolic blood pressure; M, male; MI, myocardial infarction; NNT, number needed to treat; NR, not reported; p–y, person–years of follow-up; NS, not statistically significant results; RCT, randomized controlled trial; T-Chol, total cholesterol and LDL-C, LDL cholesterol in mmol/l; wgt, weighted. aClassification in higher and lower CV risk by the presence or not of symptomatic CHD at inclusion and/or CV-risk factors as inclusion criteria; studies ordered by incidence of CHD (non-fatal and fatal MI and CHD death) event rates in control groups. bAll values correspond to population means unless otherwise indicated. cCHD events include non-fatal and fatal myocardial infarction and CHD death. dNNT calculated using the formula: {1/[Incidence rate in controls – (incidence rate in controls × risk ratio)]} /5; Risk ratios and 95% confidence intervals obtained from the meta-analysis table (Figure 1). eAll NNT calculated for an equivalent 5 years. fAverage persistence to treatment reported; subgroup summary using mean weighted by the number of p–y in the subgroup. gAdjusted NNT = unadjusted NNT × (% persistence to treatment/65%); see Methods. View Large Table 1. Characteristics of RCTs included in the analysis, NNT and persistence-adjusted NNT by CHD-risk categoriesa CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD, coronary heart disease; CVD, cardiovascular diseases; F, female; HBP, high blood pressure and SBP, systolic blood pressure; M, male; MI, myocardial infarction; NNT, number needed to treat; NR, not reported; p–y, person–years of follow-up; NS, not statistically significant results; RCT, randomized controlled trial; T-Chol, total cholesterol and LDL-C, LDL cholesterol in mmol/l; wgt, weighted. aClassification in higher and lower CV risk by the presence or not of symptomatic CHD at inclusion and/or CV-risk factors as inclusion criteria; studies ordered by incidence of CHD (non-fatal and fatal MI and CHD death) event rates in control groups. bAll values correspond to population means unless otherwise indicated. cCHD events include non-fatal and fatal myocardial infarction and CHD death. dNNT calculated using the formula: {1/[Incidence rate in controls – (incidence rate in controls × risk ratio)]} /5; Risk ratios and 95% confidence intervals obtained from the meta-analysis table (Figure 1). eAll NNT calculated for an equivalent 5 years. fAverage persistence to treatment reported; subgroup summary using mean weighted by the number of p–y in the subgroup. gAdjusted NNT = unadjusted NNT × (% persistence to treatment/65%); see Methods. View Large Five-year NNTs were calculated separately for each of the nine RCTs. Incidence rates of CHD events in control groups were calculated as the total number of CHD events divided by the number of p–y of observation in each RCT. The number of p–y of follow-up was estimated, when not reported, using the number of study subjects included and the average duration of follow-up reported. NNTs were calculated using CHD incidence rates in statin and control groups as a valid approximation of the cumulative incidence as recommended when incidence rates are low (9). NNTs were calculated for 5-year utilization of statins, which is consistent with the duration of follow-up in the RCTs (range 31 to 64 months), in compliance with the recommendation of not extrapolating NNTs beyond the actual period of observation (10). The adjustment for persistence with statin treatment in RCTs compared to a real-life population was calculated (11) by multiplying each NNT by the ratio of the average persistence rate of study subjects with their statin treatment in the RCT over a 5-year persistence of 65% derived from a cohort of 275363 new lipid-lowering-drug users (92.4% statins) followed from 2010 to 2015 through the Quebec National Drug Insurance database that covers almost all individuals over the age of 65 (12). In that cohort, persistence, defined as the number of individuals who filled a statin prescription 5 years after their initial prescription, was equal to 66.3%. This result is consistent with what is reported elsewhere (2),(7). Numbers used and details for NNT calculations are shown in Table 1 and footnotes. Data extraction and calculations were performed independently by two of the authors and discrepancies were resolved by consensus. Results Four of the nine RCTs included in this analysis were classified in the ‘lower CHD risk’ category (13–16) with CHD event rates in the control group all below 7.0 per 1000 p–y, and five in the ‘higher CHD risk’ category (17–21) with rates above 12.0 per 1000 p–y except for the PHYLLIS study (17) that included, as previously mentioned, only patients with uncontrolled or untreated hypertension (Table 1). Summary risk ratios (RR) used in the calculation of 5-year NNTs and their 95% confidence intervals (CI) stratified by the two CHD-risk categories are displayed in Figure 1. Based on I2 statistics, RR were homogeneous within but not between the two categories. Table 1 shows 5-year unadjusted and adjusted NNTs calculated for each RCT and their summary for the two CHD-risk categories. Individual adjusted NNTs were larger than their corresponding unadjusted values by 25% and 15% on average in the lower and higher CHD risk categories, respectively. Larger adjusted NNT values mean a less effective treatment, a result that is explained by the better average persistence to statin treatment in RCTs (81% and 74% in the higher and lower CHD risk categories, respectively) compared to our 5-year estimation of 65% in a real-life population. Figure 1. View largeDownload slide Meta-analysis summary table: statins and risk of non-fatal and fatal myocardial infarction and coronary heart disease (CHD) death combined—stratified by level of CHD risk. meta-analysis using Review Manager (RevMan) version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) Figure 1. View largeDownload slide Meta-analysis summary table: statins and risk of non-fatal and fatal myocardial infarction and coronary heart disease (CHD) death combined—stratified by level of CHD risk. meta-analysis using Review Manager (RevMan) version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) As expected, 5-year NNTs were larger in the lower CHD risk category. The difference was almost 3-folds with a summary value for persistence-adjusted NNT of 146 (95% CI: 117–211) (unadjusted: 117; 95% CI: 94–167) compared to the higher CHD risk category with a summary value of 53 (95% CI: 39–88) (unadjusted: 46; 95% CI: 34–78). Larger NNT values in the lower CHD risk category reflect lower summary CHD event rate in that category (3.7 per 1000 p–y) compared to the higher CHD risk category (14.4 per 1000 p–y). Discussion Our study highlights concerns about estimating NNT from studies including different CV-risk populations. In our analysis, NNTs were almost three times larger in populations at lower than at higher CHD risk because of associated CV-risk factors. In addition, it showed that the unadjusted NNT with a statin to prevent one CHD event inappropriately underestimates the number of patients to prescribe a statin. Differences between adjusted and unadjusted NNTs reflect in part the gap between persistence to statin treatment in the RCTs retained in this analysis (62% to 98%) and that observed in real-life populations. These more realistic and appropriate results regarding the use of statins in primary prevention of CHD should send a strong signal to physicians, patients and decision makers. Firstly, in absolute numbers, our results estimate that 1 out of 146 lower risk individuals (1 out of 53 of higher risk patients), who are prescribed a statin for primary prevention, experiences a CHD benefit over 5 years while no benefit is to be expected in the other 145 (52 in higher risk individuals). Thus, in the lower CHD risk populations, the perspective of benefit is very modest over a 5-year period. These figures concur to qualifying an intervention such as statin prescription in a population at low CV risk, as a pseudo-prevention strategy (22). In fact, our 95% CI for persistence-adjusted NNT values in the lower risk group (117 to 211) is well within the range of values reported in a recent review of eight meta-analyses on the number needed to harm with regard to the risk of new-onset diabetes with statins (23). Balancing this kind of harm against the benefits expected in individuals at low CV risk may lead to reconsider statin use and put more emphasis on lifestyle management instead. Secondly, our results emphasize that long-term persistence to treatment remains a challenge in spite of the best efforts. A recent systematic review showed that even in the context of randomized trials the effect of various interventions to increase adherence to statins is small at best (5). Nevertheless, our results urge health professionals to reinforce compliance to statins in their patients who are more likely to benefit from taking regularly their medication e.g. those at higher risk of cardiovascular diseases (CVD). Indeed, family physicians have a central role in inter-professional care at facilitating outreach approaches for optimizing adherence to medication and at monitoring results. Adherence to non-statin preventive options, including healthy diet, smoking cessation and regular physical activity, should also be strongly emphasized. Despite likewise challenging, these measures carry positive impacts on many other aspects of health and have better chances at improving global health and well-being in general(24, 25). Our study bears limitations. First, we limited our analysis to the RCTs selected in the Cochrane review, which included studies with <10% of patients with a history of CVD. For instance, including studies with marginally more such patients as in the ALLHAT-LLT (26) (15%) and ASCOT-LLA (27) (18%) RCTs would impact the results further away from a real primary prevention perspective. Second, the categorization of RCTs by the level of CHD risk of their study populations can be criticized as it was done post hoc. Eight of the nine RCTs could be clearly categorized on the basis of the CHD incidence rates in their control group. This criterion was not sufficient, however, to put the PHYLLIS study (17) in the higher risk category although it only included patients with uncontrolled or untreated hypertension. Nevertheless, classifying this study in one or the other category did not change our results significantly as its sample size was small compared to the others. The labels ‘lower’ and ‘higher’ CHD risk used in our classification of RCTs could roughly correspond to low (CHD event rate lower than 7% over 10 years) and moderate (CHD event rates between 12% and 16% over 10 years) CV risk using the Framingham model. This is indicative only as no direct relation can be made with the Framingham CV-risk estimation because it includes more manifestations of CVD (some at high risk of biased assessment) than the more robust definition of CHD used in this study. Future meta-analyses should plan a priori a strategy to stratify the analyses based on the baseline CHD/CV risk of the study populations included. Third, to strengthen the homogeneity of the study results and the robustness of our analysis, we chose to exclude health-related services and clinically subjective outcomes, such as hospitalizations for angina, and coronary intervention procedures, which are considered ‘soft’ outcomes (28). This was a departure from the Cochrane review with the consequence that it reduced the number of eligible RCTs entered in our analysis making the results not directly comparable to those published (1, 6). The estimation of NNTs also bears limitations. First, weighted average of NNTs had large CI due to the low event counts. Their range, however, did not overlap between the two CHD-risk categories. Second, persistence-adjusted NNT, depends for its calculation on the choice of a realistic value for persistence with statin treatment in real-life. The 65% used in our analysis is consistent with current observations in the population of the province of Quebec but does not account for incomplete adherence to prescription (80% or more of the prescriptions filled) estimated at 53% in that cohort of new statin users followed over 5 years (12). Accounting for adherence would further increase the NNT. Third, although adjusted NNT values were 15%–25% larger than their unadjusted equivalent, they still may represent an underestimation of the real number of individuals to be exposed to a statin as it does not account for other factors such as lower effectiveness of statins in new (incident) compared to former (prevalent) users (often included in RCTs) or for the healthy volunteer effect (study subjects healthier than population for equivalent risk) (29). Conclusion The estimation of crude and persistence-adjusted 5-year NNT values for statins in primary prevention of populations at lower risk of CHD is almost three times larger than those in higher risk populations. Researchers and authors should avoid reporting combined results from all RCTs conducted in primary prevention and account for the CV risk in individual studies. As recommended in most clinical practice guidelines, primary health care professionals should estimate the CV risk of their patients in order to better orient their choice of preventive approaches since not all primary prevention patients are equal. As such, and emphasized by our results, lifestyle must be the priority over statins in low CV risk individuals. Knowledge of persistence-adjusted 5-year NNT values generates more realistic expectations of outcomes and reinforces the importance of promoting and monitoring compliance in patients at higher risk who are more likely to benefit from statins. These results may also contribute to motivate governments, other third-party payers and public health organizations to promoting non-pharmacological options such as diet and exercise as primary goals setting for CV primary prevention in low-risk populations. Supplementary material Supplementary data are available at Family Practice online. Declaration Funding: No direct source of funding was received for this project that was supported in part by an in-kind contribution from the INESSS. Interpretations and opinions expressed in this manuscript are those of the authors and do not engage the INESSS Disclosure: All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation Conflict of interest: MR has participated to phase IV epidemiological studies of hypolipemiant in France conducted by LASER Analytica. No direct financial or in kind advantage was received from the sponsors of these studies, Astra-Zeneca and MSD France. PP has received honorarium for providing continuing medical education and consulting as expert from Abbott Vascular, Amgen, AstraZeneca, Boehringer Ingelheim, Bristol-Meyers Squibb, Eli Lilly, Janssen, Merck, Novartis, NovoNordisk, Pfizer, Roche, Sanofi-Aventis, Servier and Valeant. PP is a senior scholar from the Quebec Health Research Fund. No other potential conflict of interest to declare. Acknowledgements We wish to thank Professor Samy Suissa at McGill University for his judicious advice in the preparation of this manuscript. Authors are members of the Quebec National Institute for Excellence in Health and Social Services (INESSS) Working Group on Statins which also includes the participation of the following members: Bélanger MC, Centre intégré universitaire de santé et de services sociaux du Saguenay-Lac-Saint-Jean, Jonquière, Canada; Dufour C, Centre intégré universitaire de santé et de services sociaux du Saguenay-Lac-Saint-Jean, Jonquière, Canada; Gaudet-Savard T, Quebec Heart and Lung Institute, Laval University, Quebec, Canada; Juneau M, Montreal Cardiology Institute, University of Montreal, Montreal, Canada; Laroche JF, Alta Pharma, Montreal, Canada; Lejeune K, INESSS, Quebec, Canada; Perron P, Faculty of Medicine, Sherbrooke University, Sherbrooke, Canada; Prémont A, INESSS Quebec, Canada; Tardif MR, INESSS, Quebec, Canada; Vachon A, Quebec Heart and Lung Institute, Laval University, Quebec, Canada; Vanier MC, Faculty of Pharmacy, University of Montreal, Montreal, Canada. References 1. Taylor FC, Huffman M, Ebrahim S. Statin therapy for primary prevention of cardiovascular disease. JAMA  2013; 310: 2451– 2. Google Scholar CrossRef Search ADS PubMed  2. Deshpande S, Quek RG, Forbes CAet al.   A systematic review to assess adherence and persistence with statins. Curr Med Res Opin  2017; 33: 769– 78. Google Scholar CrossRef Search ADS PubMed  3. De Vera MA, Bhole V, Burns LCet al.   Impact of statin adherence on cardiovascular disease and mortality outcomes: a systematic review. Br J Clin Pharmacol  2014; 78: 684– 98. Google Scholar CrossRef Search ADS PubMed  4. van Driel ML, Morledge MD, Ulep Ret al.   Interventions to improve adherence to lipid-lowering medication. Cochrane Database Syst Rev  2016; 12: CD004371. 5. Rash JA, Campbell DJ, Tonelli Met al.   A systematic review of interventions to improve adherence to statin medication: what do we know about what works? Prev Med  2016; 90: 155– 69. Google Scholar CrossRef Search ADS PubMed  6. Taylor F, Huffman MD, Macedo AFet al.   Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev  2013; 1: CD004816. 7. Lemstra M, Blackburn D, Crawley Aet al.   Proportion and risk indicators of nonadherence to statin therapy: a meta-analysis. Can J Cardiol  2012; 28: 574– 80. Google Scholar CrossRef Search ADS PubMed  8. Redberg RF, Katz MH. Statins for primary prevention: the debate is intense, but the data are weak. JAMA  2016; 316: 1979– 81. Google Scholar CrossRef Search ADS PubMed  9. Suissa S. Number needed to treat in COPD: exacerbations versus pneumonias. Thorax  2013; 68: 540– 3. Google Scholar CrossRef Search ADS PubMed  10. Suissa S. The number needed to treat: 25 years of trials and tribulations in clinical research. Rambam Maimonides Med J  2015; 6: e0033. Google Scholar CrossRef Search ADS   11. Richard T, Vanhaeverbeek M, Van Meerhaeghe A. [The number needed to treat (NNT)]. Rev Med Brux  2011; 32: 453– 8. (French) Google Scholar PubMed  12. Institut national d’excellence en santé et en services sociaux (INESSS). [Utilization of lipid lowering drug therapies in Quebec 2010 to 2015] . Quebec, Canada: Institut national d’excellence en santé et en services sociaux (INESSS), June 2017 (in print). To be available at: https://www.inesss.qc.ca. 13. Nakamura H, Arakawa K, Itakura Het al.  ; with the MEGA Study Group. Primary prevention of cardiovascular disease with pravastatin in Japan (MEGA study): a prospective randomised controlled trial. Lancet  2006; 368: 1155– 63. Google Scholar CrossRef Search ADS PubMed  14. Ridker PM, Danielson E, Fonseca FAet al.  ; with the JUPITER study group. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med  2008; 359: 2195– 207. Google Scholar CrossRef Search ADS PubMed  15. Downs JR, Clearfield M, Weis Set al.  ; AFCAPS. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. Air force/Texas coronary atherosclerosis prevention study. JAMA  1998; 279: 1615– 22. Google Scholar CrossRef Search ADS PubMed  16. Furberg CD, Adams HPJr, Applegate WBet al.  ; ACAPS. Effect of lovastatin on early carotid atherosclerosis and cardiovascular events. Asymptomatic carotid artery progression study (ACAPS) research group. Circulation  1994; 90: 1679– 87. Google Scholar CrossRef Search ADS PubMed  17. Zanchetti A, Crepaldi G, Bond MGet al.  ; with PHYLLIS investigators. Different effects of antihypertensive regimens based on fosinopril or hydrochlorothiazide with or without lipid lowering by pravastatin on progression of asymptomatic carotid atherosclerosis: principal results of PHYLLIS—a randomized double-blind trial. Stroke  2004; 35: 2807– 12. Google Scholar CrossRef Search ADS PubMed  18. Salonen R, Nyyssönen K, Porkkala Eet al.   Kuopio atherosclerosis prevention study (KAPS): a population-based primary preventive trial of the effect of LDL lowering on atherosclerotic progression in carotid and femoral arteries. Circulation  1995; 92: 1758– 64. Google Scholar CrossRef Search ADS PubMed  19. Colhoun HM, Betteridge DJ, Durrington PNet al.  ; with CARDS investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the collaborative atorvastatin diabetes study (CARDS): multicentre randomised placebo-controlled trial. Lancet  2004; 364: 685– 96. Google Scholar CrossRef Search ADS PubMed  20. Knopp RH, d’Emden M, Smilde JGet al.  ; ASPEN. Efficacy and safety of atorvastatin in the prevention of cardiovascular end points in subjects with type 2 diabetes: the atorvastatin study for prevention of coronary heart disease endpoints in non-insulin-dependent diabetes mellitus (ASPEN). Diabetes Care  2006; 29: 1478– 85. Google Scholar CrossRef Search ADS PubMed  21. The west of Scotland coronary prevention study group. Baseline risk factors and their associations with outcome in the west of Scotland coronary prevention study. Am J Cardiol  1997; 79: 756– 62. CrossRef Search ADS PubMed  22. Chiolero A, Paradis G, Paccaud F. The pseudo-high-risk prevention strategy. Int J Epidemiol  2015; 44: 1469– 73. Google Scholar CrossRef Search ADS PubMed  23. Park ZH, Juska A, Dyakov D, Patel R. Statin-associated incident diabetes: a literature review. Consult Pharm  2014; 29: 317– 34. Google Scholar CrossRef Search ADS PubMed  24. Silverman MG, Ference BA, Im Ket al.   Association between lowering LDL-C and cardiovascular risk reduction among different therapeutic interventions: a systematic review and meta-analysis. JAMA  2016; 316: 1289– 97. Google Scholar CrossRef Search ADS PubMed  25. Zeymer U, James S, Berkenboom Get al.   with APTOR investigators. Differences in the use of guideline-recommended therapies among 14 European countries in patients with acute coronary syndromes undergoing PCI. Eur J Prev Cardiol  2013; 20: 218– 28. Google Scholar CrossRef Search ADS PubMed  26. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The antihypertensive and lipid-lowering treatment to prevent heart attack trial. Major outcomes in moderately hypercholesterolemic, hypertensive patients randomized to pravastatin vs usual care: the antihypertensive and lipid-lowering treatment to prevent heart attack trial (ALLHAT-LLT). JAMA  2002; 288: 2998– 3007. CrossRef Search ADS PubMed  27. Sever PS, Dahlöf B, Poulter NRet al.   with ASCOT investigators. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian cardiac outcomes trial-lipid lowering arm (ASCOT-LLA): a multicentre randomised controlled trial. Lancet  2003; 361: 1149– 58. Google Scholar CrossRef Search ADS PubMed  28. Thompson A, Temple NJ. The case for statins: has it really been made? J R Soc Med  2004; 97: 461– 4. Google Scholar CrossRef Search ADS PubMed  29. Danaei G, Tavakkoli M, Hernán MA. Bias in observational studies of prevalent users: lessons for comparative effectiveness research from a meta-analysis of statins. Am J Epidemiol  2012; 175: 250– 62. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Family Practice Oxford University Press

Number of patients needed to prescribe statins in primary cardiovascular prevention: mirage and reality

Loading next page...
 
/lp/ou_press/number-of-patients-needed-to-prescribe-statins-in-primary-Ej0a3dXktQ
Publisher
Oxford University Press
Copyright
© The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0263-2136
eISSN
1460-2229
D.O.I.
10.1093/fampra/cmx124
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Number of patients needed to treat (NNT) with a statin in primary prevention of coronary heart disease (CHD) is often misinterpreted because this single statistic averages results from heterogeneous studies. Objective To provide estimates of the number of individuals needed to be prescribed a statin to prevent one CHD event accounting for their level of CHD risk and for persistence to treatment. Methods A post hoc analysis was conducted based on a Cochrane review on statins for the primary prevention of cardiovascular diseases. Five-year NNTs were calculated separately from randomized clinical trials (RCTs), including ‘lower’ and ‘higher’ risk populations (CHD mean event rates of 3.7 and 14.4 per 1000 person–years, respectively). NNTs were adjusted for 5-year persistence to treatment using a value of 65%. Results Persistence-adjusted 5-year NNTs to prevent one CHD for the lower and higher CHD risk categories were 146 [95% confidence interval (CI): 117–211] and 53 (95% CI: 39–88) respectively, values 25% and 15% higher than their unadjusted counterpart (117, 95% CI: 94–167 and 46, 95% CI: 34–78). Conclusions Five-year NNTs for statins to prevent a first CHD is almost three times higher in those at lower versus higher risk populations. Reporting combined results from RCTs including subjects at different cardiovascular risks should be avoided. Individualizing the risk of CHD should orient family physicians and their patients in their choice of preventive approaches and generate more realistic expectations about compliance and outcomes. Adherence persistence, coronary artery disease, statins primary prevention, number needed to treat Introduction Number needed to treat (NNT) is often used in medical publications to translate complex research results into a simple metric easy to interpret and use in medical decision making. For instance, a 5-year NNT of 88 is interpreted as the estimated number of patients needed to be treated with statins over 5 years to prevent the occurrence of a first coronary heart disease (CHD) event (1). Although generally acknowledged, the actual impact of lower persistence to statin treatment in real-life compared with research populations from which NNT values are derived has not been adequately accounted for in NNT estimations (2–3). As the effectiveness of interventions to improve persistence to statins has been shown to be modest, it is imperative that clinicians, patients and decision makers have access to the right information, so they can base their treatment and health policy decisions on realistic expectations (4–5). We hypothesize that an NNT of 88 derived from meta-analysis of randomized clinical trials (RCTs) of statins for primary cardiovascular (CV) prevention does not reflect the true number of individuals needed to be prescribed a statin to prevent a first CHD event, because it includes populations at different CV risks with diverse outcomes, and because the drop-out rate from treatment is higher in real-life than in RCTs (6–7). The objective of this study was to provide estimates of the 5-year number of individuals needed to be prescribed a statin to prevent a first non-fatal or fatal myocardial infarction (MI) or CHD death accounting for their level of CHD risk and for real-life persistence to statin treatment in primary care. Methods This post hoc analysis used the results of nine RCTs included in a Cochrane meta-analysis of statins for the primary prevention of CV disease, that specifically reported outcomes on fatal and non-fatal MI and CHD deaths (supplementary Table) (6). CHD outcomes based on symptoms (angina and unstable angina) and coronary revascularization were excluded from this analysis as they may not have been uniformly defined and systematically reported between RCTs (i.e. ‘soft’ outcomes) (8). The nine RCTs were classified in two broad CHD-risk categories labelled ‘lower’ and ‘higher’, primarily based on the ranking of CHD event rates observed in the control group of each RCT. The ranking showed a gap between the ACAPS study with an event rate of 6.9 per 1000 person–years (p–y) and the KAPS study with a rate of 12.6 per 1000 p–y, which was used to delineate our two categories (Table 1). The PHYLLIS study (event rate of 4.6 per 1000 p–y) was reclassified in the higher CHD risk category because it included only patients with uncontrolled or untreated hypertension. The number of CHD events (fatal and non-fatal MI, and CHD deaths) in statin and control groups for each RCT were entered in a meta-analysis table stratified by the two CHD risk categories using Review Manager (RevMan) version 5.3 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). Table 1. Characteristics of RCTs included in the analysis, NNT and persistence-adjusted NNT by CHD-risk categoriesa CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD, coronary heart disease; CVD, cardiovascular diseases; F, female; HBP, high blood pressure and SBP, systolic blood pressure; M, male; MI, myocardial infarction; NNT, number needed to treat; NR, not reported; p–y, person–years of follow-up; NS, not statistically significant results; RCT, randomized controlled trial; T-Chol, total cholesterol and LDL-C, LDL cholesterol in mmol/l; wgt, weighted. aClassification in higher and lower CV risk by the presence or not of symptomatic CHD at inclusion and/or CV-risk factors as inclusion criteria; studies ordered by incidence of CHD (non-fatal and fatal MI and CHD death) event rates in control groups. bAll values correspond to population means unless otherwise indicated. cCHD events include non-fatal and fatal myocardial infarction and CHD death. dNNT calculated using the formula: {1/[Incidence rate in controls – (incidence rate in controls × risk ratio)]} /5; Risk ratios and 95% confidence intervals obtained from the meta-analysis table (Figure 1). eAll NNT calculated for an equivalent 5 years. fAverage persistence to treatment reported; subgroup summary using mean weighted by the number of p–y in the subgroup. gAdjusted NNT = unadjusted NNT × (% persistence to treatment/65%); see Methods. View Large Table 1. Characteristics of RCTs included in the analysis, NNT and persistence-adjusted NNT by CHD-risk categoriesa CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD grouping  Study Population Years of inclusion (Reference)  Average follow-up (Planned maximum) in months  n (total) % Males Mean age (Range)  Population characteristics at inclusionb  Incidence rate CHD eventsc Controls per 1000 p–y (n/p–y)  NNTd Eq. 5 yearse  Persistencef to treatment  Persistence- adjusted NNTg Eq. 5 yearse  Adjusted– unadjusted NNT Absolute (relative) difference  Lower CHD Risk  MEGA, Japan, 1994–1999 (13)  63.6 (60)  n = 7832 32% 59 (40–70)  Current and past smoking: 20.5% T-Chol: 6.3; LDL-C: 4.1 HBP (controlled only): 42%; SBP: 132 Diabetes (controlled only): 21% History of CVD: none  2.1 43/21020  208 (145–834)  90%  288 (201–1154)  +80 (+38%)    JUPITER, 26 countries, 2003–2006 (14)  22.8 (60)  n = 17802 62% 66 (M > 50; F > 60)  Current smoking: 16% T-Chol (median): 4.8; LDL-C (median): 2.8 HBP (controlled only): NR; SBP: 134 Diabetes: none History of CVD: none  3.7 68/18455  100 (78–180)  75%  115 (90–208)  +15 (+15%)    AFCAPS/TexCAPS, USA, 1990–1993 (15)  62.4 (NR)  n = 6606 58% M: 58 (45–73) F: 63 (55–73)  Current smoking: 12.5% T-Chol: 5.7; LDL-C: 3.9 HBP (controlled only) 22%; SBP: 138 Diabetes (type 2 controlled only): 2.5% History of CVD: none  5.7 95/16797  88 (62–197)  75%  102 (71–227)  +14 (+16%)    ACAPS, USA, 1989–1990 (16)  34.1 (36)  n = 919 52% 62 (40–79)  Current smoking: 11.9% T-Chol: 6.1; LDL-C: 4.0 HBP: 29%; SBP: 131 Diabetes type 2: 2.3% History of CVD: 100% asymptomatic carotid atherosclerosis  6.9 9/1304  65 (NS)  85%  85 (NS)  +20 (+31%)    Lower CHD risk Summary        3.7 215/57576  117 (94–167)  81% (wgt.average)  146 (117–211)  +29 (+25%)  Higher CHD risk  PHYLLIS, Italy, < 2004 (17)  31.2 (36)  n = 253 41% 58 (45–70)  Current smoking: 20% T-Chol: 6.8; LDL-C: 4.7 HBP uncontrolled or untreated: 100%; SBP: 160 Diabetes: NR History of CVD: 100% asymptomatic carotid atherosclerosis  4.6 3/660  66 (NS)  98%  100 (NS)  +34 (+52%)    KAPS, Finland, 1990 (18)  NR (36)  n = 447 100% 57 (44–65)  Current smoking: 26% T-Chol: 6.7; LDL-C: 4.9 HBP: 33%; SBP: 136 Diabetes (not specified): 2.5% History of CVD: MI 6% to 9%  12.6 8/636  42 (NS)  93%  59 (NS)  +17 (+41%)    CARDS, U.K. and Ireland, 1997–2001 (19)  46.8 (60)  n = 2838 68% 62 (40–75)  Current smoking: 22.5% T-Chol: 5.4; LDL-C: 3.0 HBP: 84%; SBP: 144 Diabetes (type 2): 100% History of CVD: none  12.6 65/5166  46 (29–318)  85%  61 (38–415)  +15 (+33%)    ASPEN, 14 countries, 1996–1999 (20)  48.0 (51)  n = 2410 63% 61 (40–75)  Current smoking: 13% T-Chol: 5.0; LDL-C: 2.9 HBP (controlled only): 53%; SBP: 133 Diabetes (type 2): 100% History of CVD: angina: 6%  15.0 34/2270  70 (NS)  62%  67 (NS)  –3 (–4%)    WOSCOPS, Scotland, 1989–1991 (21)  58.8 (60)  n = 6595 100% 55 (45–64)  Current smoking: 44% T-Chol: 7.0; LDL-C: 5.0 HBP: 16%; SBP: 136 Diabetes (not specified): 1% History of CVD: angina: 5%  15.4 248/16136  43 (31–82)  70%  59 (34–88)  +16 (+37%)    Higher CHD risk Summary         14.4 358/24868  46 (34–78)  74% (wgt. average)  53 (39–88)  +7 (+15%)  CHD, coronary heart disease; CVD, cardiovascular diseases; F, female; HBP, high blood pressure and SBP, systolic blood pressure; M, male; MI, myocardial infarction; NNT, number needed to treat; NR, not reported; p–y, person–years of follow-up; NS, not statistically significant results; RCT, randomized controlled trial; T-Chol, total cholesterol and LDL-C, LDL cholesterol in mmol/l; wgt, weighted. aClassification in higher and lower CV risk by the presence or not of symptomatic CHD at inclusion and/or CV-risk factors as inclusion criteria; studies ordered by incidence of CHD (non-fatal and fatal MI and CHD death) event rates in control groups. bAll values correspond to population means unless otherwise indicated. cCHD events include non-fatal and fatal myocardial infarction and CHD death. dNNT calculated using the formula: {1/[Incidence rate in controls – (incidence rate in controls × risk ratio)]} /5; Risk ratios and 95% confidence intervals obtained from the meta-analysis table (Figure 1). eAll NNT calculated for an equivalent 5 years. fAverage persistence to treatment reported; subgroup summary using mean weighted by the number of p–y in the subgroup. gAdjusted NNT = unadjusted NNT × (% persistence to treatment/65%); see Methods. View Large Five-year NNTs were calculated separately for each of the nine RCTs. Incidence rates of CHD events in control groups were calculated as the total number of CHD events divided by the number of p–y of observation in each RCT. The number of p–y of follow-up was estimated, when not reported, using the number of study subjects included and the average duration of follow-up reported. NNTs were calculated using CHD incidence rates in statin and control groups as a valid approximation of the cumulative incidence as recommended when incidence rates are low (9). NNTs were calculated for 5-year utilization of statins, which is consistent with the duration of follow-up in the RCTs (range 31 to 64 months), in compliance with the recommendation of not extrapolating NNTs beyond the actual period of observation (10). The adjustment for persistence with statin treatment in RCTs compared to a real-life population was calculated (11) by multiplying each NNT by the ratio of the average persistence rate of study subjects with their statin treatment in the RCT over a 5-year persistence of 65% derived from a cohort of 275363 new lipid-lowering-drug users (92.4% statins) followed from 2010 to 2015 through the Quebec National Drug Insurance database that covers almost all individuals over the age of 65 (12). In that cohort, persistence, defined as the number of individuals who filled a statin prescription 5 years after their initial prescription, was equal to 66.3%. This result is consistent with what is reported elsewhere (2),(7). Numbers used and details for NNT calculations are shown in Table 1 and footnotes. Data extraction and calculations were performed independently by two of the authors and discrepancies were resolved by consensus. Results Four of the nine RCTs included in this analysis were classified in the ‘lower CHD risk’ category (13–16) with CHD event rates in the control group all below 7.0 per 1000 p–y, and five in the ‘higher CHD risk’ category (17–21) with rates above 12.0 per 1000 p–y except for the PHYLLIS study (17) that included, as previously mentioned, only patients with uncontrolled or untreated hypertension (Table 1). Summary risk ratios (RR) used in the calculation of 5-year NNTs and their 95% confidence intervals (CI) stratified by the two CHD-risk categories are displayed in Figure 1. Based on I2 statistics, RR were homogeneous within but not between the two categories. Table 1 shows 5-year unadjusted and adjusted NNTs calculated for each RCT and their summary for the two CHD-risk categories. Individual adjusted NNTs were larger than their corresponding unadjusted values by 25% and 15% on average in the lower and higher CHD risk categories, respectively. Larger adjusted NNT values mean a less effective treatment, a result that is explained by the better average persistence to statin treatment in RCTs (81% and 74% in the higher and lower CHD risk categories, respectively) compared to our 5-year estimation of 65% in a real-life population. Figure 1. View largeDownload slide Meta-analysis summary table: statins and risk of non-fatal and fatal myocardial infarction and coronary heart disease (CHD) death combined—stratified by level of CHD risk. meta-analysis using Review Manager (RevMan) version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) Figure 1. View largeDownload slide Meta-analysis summary table: statins and risk of non-fatal and fatal myocardial infarction and coronary heart disease (CHD) death combined—stratified by level of CHD risk. meta-analysis using Review Manager (RevMan) version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) As expected, 5-year NNTs were larger in the lower CHD risk category. The difference was almost 3-folds with a summary value for persistence-adjusted NNT of 146 (95% CI: 117–211) (unadjusted: 117; 95% CI: 94–167) compared to the higher CHD risk category with a summary value of 53 (95% CI: 39–88) (unadjusted: 46; 95% CI: 34–78). Larger NNT values in the lower CHD risk category reflect lower summary CHD event rate in that category (3.7 per 1000 p–y) compared to the higher CHD risk category (14.4 per 1000 p–y). Discussion Our study highlights concerns about estimating NNT from studies including different CV-risk populations. In our analysis, NNTs were almost three times larger in populations at lower than at higher CHD risk because of associated CV-risk factors. In addition, it showed that the unadjusted NNT with a statin to prevent one CHD event inappropriately underestimates the number of patients to prescribe a statin. Differences between adjusted and unadjusted NNTs reflect in part the gap between persistence to statin treatment in the RCTs retained in this analysis (62% to 98%) and that observed in real-life populations. These more realistic and appropriate results regarding the use of statins in primary prevention of CHD should send a strong signal to physicians, patients and decision makers. Firstly, in absolute numbers, our results estimate that 1 out of 146 lower risk individuals (1 out of 53 of higher risk patients), who are prescribed a statin for primary prevention, experiences a CHD benefit over 5 years while no benefit is to be expected in the other 145 (52 in higher risk individuals). Thus, in the lower CHD risk populations, the perspective of benefit is very modest over a 5-year period. These figures concur to qualifying an intervention such as statin prescription in a population at low CV risk, as a pseudo-prevention strategy (22). In fact, our 95% CI for persistence-adjusted NNT values in the lower risk group (117 to 211) is well within the range of values reported in a recent review of eight meta-analyses on the number needed to harm with regard to the risk of new-onset diabetes with statins (23). Balancing this kind of harm against the benefits expected in individuals at low CV risk may lead to reconsider statin use and put more emphasis on lifestyle management instead. Secondly, our results emphasize that long-term persistence to treatment remains a challenge in spite of the best efforts. A recent systematic review showed that even in the context of randomized trials the effect of various interventions to increase adherence to statins is small at best (5). Nevertheless, our results urge health professionals to reinforce compliance to statins in their patients who are more likely to benefit from taking regularly their medication e.g. those at higher risk of cardiovascular diseases (CVD). Indeed, family physicians have a central role in inter-professional care at facilitating outreach approaches for optimizing adherence to medication and at monitoring results. Adherence to non-statin preventive options, including healthy diet, smoking cessation and regular physical activity, should also be strongly emphasized. Despite likewise challenging, these measures carry positive impacts on many other aspects of health and have better chances at improving global health and well-being in general(24, 25). Our study bears limitations. First, we limited our analysis to the RCTs selected in the Cochrane review, which included studies with <10% of patients with a history of CVD. For instance, including studies with marginally more such patients as in the ALLHAT-LLT (26) (15%) and ASCOT-LLA (27) (18%) RCTs would impact the results further away from a real primary prevention perspective. Second, the categorization of RCTs by the level of CHD risk of their study populations can be criticized as it was done post hoc. Eight of the nine RCTs could be clearly categorized on the basis of the CHD incidence rates in their control group. This criterion was not sufficient, however, to put the PHYLLIS study (17) in the higher risk category although it only included patients with uncontrolled or untreated hypertension. Nevertheless, classifying this study in one or the other category did not change our results significantly as its sample size was small compared to the others. The labels ‘lower’ and ‘higher’ CHD risk used in our classification of RCTs could roughly correspond to low (CHD event rate lower than 7% over 10 years) and moderate (CHD event rates between 12% and 16% over 10 years) CV risk using the Framingham model. This is indicative only as no direct relation can be made with the Framingham CV-risk estimation because it includes more manifestations of CVD (some at high risk of biased assessment) than the more robust definition of CHD used in this study. Future meta-analyses should plan a priori a strategy to stratify the analyses based on the baseline CHD/CV risk of the study populations included. Third, to strengthen the homogeneity of the study results and the robustness of our analysis, we chose to exclude health-related services and clinically subjective outcomes, such as hospitalizations for angina, and coronary intervention procedures, which are considered ‘soft’ outcomes (28). This was a departure from the Cochrane review with the consequence that it reduced the number of eligible RCTs entered in our analysis making the results not directly comparable to those published (1, 6). The estimation of NNTs also bears limitations. First, weighted average of NNTs had large CI due to the low event counts. Their range, however, did not overlap between the two CHD-risk categories. Second, persistence-adjusted NNT, depends for its calculation on the choice of a realistic value for persistence with statin treatment in real-life. The 65% used in our analysis is consistent with current observations in the population of the province of Quebec but does not account for incomplete adherence to prescription (80% or more of the prescriptions filled) estimated at 53% in that cohort of new statin users followed over 5 years (12). Accounting for adherence would further increase the NNT. Third, although adjusted NNT values were 15%–25% larger than their unadjusted equivalent, they still may represent an underestimation of the real number of individuals to be exposed to a statin as it does not account for other factors such as lower effectiveness of statins in new (incident) compared to former (prevalent) users (often included in RCTs) or for the healthy volunteer effect (study subjects healthier than population for equivalent risk) (29). Conclusion The estimation of crude and persistence-adjusted 5-year NNT values for statins in primary prevention of populations at lower risk of CHD is almost three times larger than those in higher risk populations. Researchers and authors should avoid reporting combined results from all RCTs conducted in primary prevention and account for the CV risk in individual studies. As recommended in most clinical practice guidelines, primary health care professionals should estimate the CV risk of their patients in order to better orient their choice of preventive approaches since not all primary prevention patients are equal. As such, and emphasized by our results, lifestyle must be the priority over statins in low CV risk individuals. Knowledge of persistence-adjusted 5-year NNT values generates more realistic expectations of outcomes and reinforces the importance of promoting and monitoring compliance in patients at higher risk who are more likely to benefit from statins. These results may also contribute to motivate governments, other third-party payers and public health organizations to promoting non-pharmacological options such as diet and exercise as primary goals setting for CV primary prevention in low-risk populations. Supplementary material Supplementary data are available at Family Practice online. Declaration Funding: No direct source of funding was received for this project that was supported in part by an in-kind contribution from the INESSS. Interpretations and opinions expressed in this manuscript are those of the authors and do not engage the INESSS Disclosure: All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation Conflict of interest: MR has participated to phase IV epidemiological studies of hypolipemiant in France conducted by LASER Analytica. No direct financial or in kind advantage was received from the sponsors of these studies, Astra-Zeneca and MSD France. PP has received honorarium for providing continuing medical education and consulting as expert from Abbott Vascular, Amgen, AstraZeneca, Boehringer Ingelheim, Bristol-Meyers Squibb, Eli Lilly, Janssen, Merck, Novartis, NovoNordisk, Pfizer, Roche, Sanofi-Aventis, Servier and Valeant. PP is a senior scholar from the Quebec Health Research Fund. No other potential conflict of interest to declare. Acknowledgements We wish to thank Professor Samy Suissa at McGill University for his judicious advice in the preparation of this manuscript. Authors are members of the Quebec National Institute for Excellence in Health and Social Services (INESSS) Working Group on Statins which also includes the participation of the following members: Bélanger MC, Centre intégré universitaire de santé et de services sociaux du Saguenay-Lac-Saint-Jean, Jonquière, Canada; Dufour C, Centre intégré universitaire de santé et de services sociaux du Saguenay-Lac-Saint-Jean, Jonquière, Canada; Gaudet-Savard T, Quebec Heart and Lung Institute, Laval University, Quebec, Canada; Juneau M, Montreal Cardiology Institute, University of Montreal, Montreal, Canada; Laroche JF, Alta Pharma, Montreal, Canada; Lejeune K, INESSS, Quebec, Canada; Perron P, Faculty of Medicine, Sherbrooke University, Sherbrooke, Canada; Prémont A, INESSS Quebec, Canada; Tardif MR, INESSS, Quebec, Canada; Vachon A, Quebec Heart and Lung Institute, Laval University, Quebec, Canada; Vanier MC, Faculty of Pharmacy, University of Montreal, Montreal, Canada. References 1. Taylor FC, Huffman M, Ebrahim S. Statin therapy for primary prevention of cardiovascular disease. JAMA  2013; 310: 2451– 2. Google Scholar CrossRef Search ADS PubMed  2. Deshpande S, Quek RG, Forbes CAet al.   A systematic review to assess adherence and persistence with statins. Curr Med Res Opin  2017; 33: 769– 78. Google Scholar CrossRef Search ADS PubMed  3. De Vera MA, Bhole V, Burns LCet al.   Impact of statin adherence on cardiovascular disease and mortality outcomes: a systematic review. Br J Clin Pharmacol  2014; 78: 684– 98. Google Scholar CrossRef Search ADS PubMed  4. van Driel ML, Morledge MD, Ulep Ret al.   Interventions to improve adherence to lipid-lowering medication. Cochrane Database Syst Rev  2016; 12: CD004371. 5. Rash JA, Campbell DJ, Tonelli Met al.   A systematic review of interventions to improve adherence to statin medication: what do we know about what works? Prev Med  2016; 90: 155– 69. Google Scholar CrossRef Search ADS PubMed  6. Taylor F, Huffman MD, Macedo AFet al.   Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev  2013; 1: CD004816. 7. Lemstra M, Blackburn D, Crawley Aet al.   Proportion and risk indicators of nonadherence to statin therapy: a meta-analysis. Can J Cardiol  2012; 28: 574– 80. Google Scholar CrossRef Search ADS PubMed  8. Redberg RF, Katz MH. Statins for primary prevention: the debate is intense, but the data are weak. JAMA  2016; 316: 1979– 81. Google Scholar CrossRef Search ADS PubMed  9. Suissa S. Number needed to treat in COPD: exacerbations versus pneumonias. Thorax  2013; 68: 540– 3. Google Scholar CrossRef Search ADS PubMed  10. Suissa S. The number needed to treat: 25 years of trials and tribulations in clinical research. Rambam Maimonides Med J  2015; 6: e0033. Google Scholar CrossRef Search ADS   11. Richard T, Vanhaeverbeek M, Van Meerhaeghe A. [The number needed to treat (NNT)]. Rev Med Brux  2011; 32: 453– 8. (French) Google Scholar PubMed  12. Institut national d’excellence en santé et en services sociaux (INESSS). [Utilization of lipid lowering drug therapies in Quebec 2010 to 2015] . Quebec, Canada: Institut national d’excellence en santé et en services sociaux (INESSS), June 2017 (in print). To be available at: https://www.inesss.qc.ca. 13. Nakamura H, Arakawa K, Itakura Het al.  ; with the MEGA Study Group. Primary prevention of cardiovascular disease with pravastatin in Japan (MEGA study): a prospective randomised controlled trial. Lancet  2006; 368: 1155– 63. Google Scholar CrossRef Search ADS PubMed  14. Ridker PM, Danielson E, Fonseca FAet al.  ; with the JUPITER study group. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med  2008; 359: 2195– 207. Google Scholar CrossRef Search ADS PubMed  15. Downs JR, Clearfield M, Weis Set al.  ; AFCAPS. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. Air force/Texas coronary atherosclerosis prevention study. JAMA  1998; 279: 1615– 22. Google Scholar CrossRef Search ADS PubMed  16. Furberg CD, Adams HPJr, Applegate WBet al.  ; ACAPS. Effect of lovastatin on early carotid atherosclerosis and cardiovascular events. Asymptomatic carotid artery progression study (ACAPS) research group. Circulation  1994; 90: 1679– 87. Google Scholar CrossRef Search ADS PubMed  17. Zanchetti A, Crepaldi G, Bond MGet al.  ; with PHYLLIS investigators. Different effects of antihypertensive regimens based on fosinopril or hydrochlorothiazide with or without lipid lowering by pravastatin on progression of asymptomatic carotid atherosclerosis: principal results of PHYLLIS—a randomized double-blind trial. Stroke  2004; 35: 2807– 12. Google Scholar CrossRef Search ADS PubMed  18. Salonen R, Nyyssönen K, Porkkala Eet al.   Kuopio atherosclerosis prevention study (KAPS): a population-based primary preventive trial of the effect of LDL lowering on atherosclerotic progression in carotid and femoral arteries. Circulation  1995; 92: 1758– 64. Google Scholar CrossRef Search ADS PubMed  19. Colhoun HM, Betteridge DJ, Durrington PNet al.  ; with CARDS investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the collaborative atorvastatin diabetes study (CARDS): multicentre randomised placebo-controlled trial. Lancet  2004; 364: 685– 96. Google Scholar CrossRef Search ADS PubMed  20. Knopp RH, d’Emden M, Smilde JGet al.  ; ASPEN. Efficacy and safety of atorvastatin in the prevention of cardiovascular end points in subjects with type 2 diabetes: the atorvastatin study for prevention of coronary heart disease endpoints in non-insulin-dependent diabetes mellitus (ASPEN). Diabetes Care  2006; 29: 1478– 85. Google Scholar CrossRef Search ADS PubMed  21. The west of Scotland coronary prevention study group. Baseline risk factors and their associations with outcome in the west of Scotland coronary prevention study. Am J Cardiol  1997; 79: 756– 62. CrossRef Search ADS PubMed  22. Chiolero A, Paradis G, Paccaud F. The pseudo-high-risk prevention strategy. Int J Epidemiol  2015; 44: 1469– 73. Google Scholar CrossRef Search ADS PubMed  23. Park ZH, Juska A, Dyakov D, Patel R. Statin-associated incident diabetes: a literature review. Consult Pharm  2014; 29: 317– 34. Google Scholar CrossRef Search ADS PubMed  24. Silverman MG, Ference BA, Im Ket al.   Association between lowering LDL-C and cardiovascular risk reduction among different therapeutic interventions: a systematic review and meta-analysis. JAMA  2016; 316: 1289– 97. Google Scholar CrossRef Search ADS PubMed  25. Zeymer U, James S, Berkenboom Get al.   with APTOR investigators. Differences in the use of guideline-recommended therapies among 14 European countries in patients with acute coronary syndromes undergoing PCI. Eur J Prev Cardiol  2013; 20: 218– 28. Google Scholar CrossRef Search ADS PubMed  26. ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The antihypertensive and lipid-lowering treatment to prevent heart attack trial. Major outcomes in moderately hypercholesterolemic, hypertensive patients randomized to pravastatin vs usual care: the antihypertensive and lipid-lowering treatment to prevent heart attack trial (ALLHAT-LLT). JAMA  2002; 288: 2998– 3007. CrossRef Search ADS PubMed  27. Sever PS, Dahlöf B, Poulter NRet al.   with ASCOT investigators. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian cardiac outcomes trial-lipid lowering arm (ASCOT-LLA): a multicentre randomised controlled trial. Lancet  2003; 361: 1149– 58. Google Scholar CrossRef Search ADS PubMed  28. Thompson A, Temple NJ. The case for statins: has it really been made? J R Soc Med  2004; 97: 461– 4. Google Scholar CrossRef Search ADS PubMed  29. Danaei G, Tavakkoli M, Hernán MA. Bias in observational studies of prevalent users: lessons for comparative effectiveness research from a meta-analysis of statins. Am J Epidemiol  2012; 175: 250– 62. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Family PracticeOxford University Press

Published: Dec 18, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off