Correspondence
between individuals at baseline will start to have possible effects on the results. The conclusion that lifestyle interventions for people with impaired glucose tolerance lowers diabetesrelated deaths, and that these effects are true in women especially, thus seem somewhat premature. I declare no competing interests.
Mattias Brunström [email protected] Department of Public Health and Clinical Medicine, Division of Medicine, Umea University, Umea 901 87-SE, Sweden 1
2
3
4
Li G, Zhang P, Wang J, et al. Cardiovascular mortality, all-cause mortality, and diabetes incidence after lifestyle intervention for people with impaired glucose tolerance in the Da Qing Diabetes Prevention Study: a 23-year follow-up study. Lancet Diabetes Endocrinol 2014; 2: 474–80. Bassler D, Montori VM, Briel M, et al. Reflections on meta-analyses involving trials stopped early for benefit: Is there a problem and if so, what is it? Stat Methods Med Res 2013; 22: 159–68. Schulz KF, Grimes DA. Multiplicity in randomised trials II: subgroup and interim analyses. Lancet 2005; 365: 1657–61. Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle interventions to prevent diabetes in the China Da Qing Diabetes Prevention Study: a 20-year follow-up study. Lancet 2008; 371: 1783–89.
Authors’ reply We thank Mattias Bronström for his interest in our Article.1 The Da Qing Diabetes Prevention Study was not terminated prematurely, but ran the planned full course of 6 years of intervention.2 Subsequently, 20 years after the initial randomisation, we designed and implemented an observational follow-up study with prespecified outcomes and reported the results on the macrocomplications and microcomplications of diabetes.3,4 Because of potentially very important, but non-significant, trends in mortality seen in the 20 year followup, a subsequent follow-up at 23 years was specifically designed to determine whether the death rates between the lifestyle intervention and control groups continued to diverge.1 The observation of a greater effect of lifestyle intervention in women, which was the result of a post-hoc 540
analysis, was not anticipated and the explanations remain unclear. Given its level of statistical significance, we believe that this result needed to be reported. However, we do share some of Bronström’s hesitation to accept its magnitude at face value. More importantly, as the first report of a randomised controlled trial to show statistically significant reduction in mortality rates after lifestyle intervention in patients with impaired glucose tolerance,1 replication of the findings from continuing studies such as The Diabetes Prevention Program Outcomes Study (DPPOS)5 will be important to determine the extent to which our results can be generalised. We declare no competing interests.
*Guangwei Li, Ping Zhang, Edward W Gregg, Michael M Engelgau, Peter H Bennett [email protected] Department of Endocrinology, China–Japan Friendship Hospital, Beijing 100029, China (GL); Center of Endocrinology and Cardiovascular Disease, National Center of Cardiology and Fuwai Hospital, Beijing, China (GL); Division of Diabetes Translation (PZ, EWG) and Center for Global Health (MME), Centers for Disease Control and Prevention, Atlanta, GA 30333, USA; and Phoenix Epidemiology and Clinical Research Branch, National Institute of Diabetes and Digestive and Kidney Diseases, Phoenix, AZ, USA (PHB) 1
2
3
4
5
Li G, Zhang P, Wang J, et al. Cardiovascular mortality, all-cause mortality, and diabetes incidence after lifestyle intervention for people with impaired glucose tolerance in the Da Qing Diabetes Prevention Study: a 23-year follow-up study. Lancet Diabetes Endocrinol 2014; 2: 474–80. Pan XR, Li GW, Hu YH, et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: 537–44. Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle interventions to prevent diabetes in the China Da Qing Diabetes Prevention Study: a 20-year follow-up study. Lancet 2008; 371: 1783–89. Gong Q, Gregg EW, Wang J, et al. Long-term effects of a randomised trial of a 6-year lifestyle intervention in impaired glucose tolerance on diabetes-related microvascular complications: the China Da Qing Diabetes Prevention Outcome Study. Diabetologia 2011; 54: 300–07. Knowler WC, Fowler SE, Hamman RF, et al. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86.
Vitamin D and falls In their Article in The Lancet Diabetes & Endocrinology (April 24), Bolland and colleagues1 conclude that “at present, there is little justification for prescribing vitamin D supplements to prevent falls”. The authors’ key underpinning assumption rests on their belief that a threshold risk reduction of 15% “is the smallest effect that is clinically relevant for an individual”. In our view, this statement naturally warrants examination in context. We are carrying out an economic evaluation of alternative public health strategies with regards to vitamin D supplementation. A simple 5 year budget impact model using UK inputs for outcomes2 and costs, taking into account institutional care for older adults,3 suggests that even if the effect of vitamin D supplementation (800 IU every day) on falls was as little as a 6% relative risk reduction, use of this intervention in individuals older than 75 years would save costs for the UK National Health Service. In this conservative scenario, more than 6000 deaths would be avoided and a further 7000 older adults could remain living independently (with a higher quality of life). Usefully, the work of Bolland’s group in Auckland does show the substantial heterogeneity between the few trials available. Some specific subgroups might benefit, but traditional aggregate data meta-analysis is poorly placed to detect these benefits reliably. The statistical power of aggregate data meta-analysis in simulation studies4 can be dramatically and consistently lower than that in equivalent individual patient data meta-analyses, with little agreement between the estimates obtained from the two methods. Only in meta-analyses of large numbers of large trials does meta-regression detect differential treatment effects between risk groups with any consistency. Therefore, while meta-analysis of summary data might be adequate to estimate a single pooled treatment
www.thelancet.com/diabetes-endocrinology Vol 2 July 2014
Correspondence
effect or to investigate study-level characteristics, when investigators are assessing whether patient characteristics are related to treatment, individual patient data analysis will generally be necessary to discover any such relations. In this regard, we agree with the authors that widespread, indiscriminate, supplementation with vitamin D has a poor evidence base, but disagree with their arbitrary threshold of clinical relevance. We suggest that to determine which patients to target with treatment, more trials of consistent design are needed. In the meantime, the authors should heed their own advice regarding meta-analyses of vitamin D and “stop torturing these data”5. JSD declares consultancy work and speaker fees for Prostrakan, Internis Pharmaceuticals, and Sandoz.
*J Stephen Davies, Chris D Poole [email protected] University Hospital of Wales, Cardiff CF14 4XW, UK (JSD) and Cochrane Institute for Primary Care and Public Health, Cardiff University, Cardiff, UK (CDP) 1
2
3
4
5
Bolland MJ, Grey A, Gamble GD, Reid IR. Vitamin D supplementation and falls: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; published online April 24. http://dx.doi. org/10.1016/S2213-8587(14)70068-3. Scuffham P, Chaplin S LR. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003; 57: 740–44. Bebbington A, Darton R, Netten A. Elderly people admitted to residential and nursing homes: 42 months on. PSSRU. November, 2000. http://www.pssru.ac.uk/pdf/p051.pdf (accessed June 9, 2014). Lambert PC, Sutton AJ, Abrams KR, Jones DR. A comparison of summary patient-level covariates in meta-regression with individual patient data meta-analysis. J Clin Epidemiol 2002; 55: 86–94. Grey A, Bolland M, Reid I. Vitamin D dose requirements for fracture prevention. N Engl J Med 2012; 367: 1367–70.
Authors’ reply In their Correspondence, Davies and Poole estimate that a 6% reduction in fall risk associated with vitamin D supplementation would be cost effective. This calculation might be correct, but conflating cost effectiveness for the population with individual clinical benefit is unwise. The absolute benefit of a 6% relative risk reduction for individuals is very small. For a group with a 10% per year risk of falls, about 27 of 100 people will fall in 3 years. If the entire group took a treatment that reduced the risk by 6%, 25 people will fall by 3 years— ie, of 100 people, two benefit and 98 do not. It seems unlikely that people would take a drug regularly for 3 years for such a small benefit, only because it is regarded as cost-effective. Furthermore, proving that vitamin D reduces falls by 6% would need very large trials or meta-analyses: the optimum sample size for a trial sequential analysis that uses this threshold is roughly 57 000 people. Notably, trial sequential analysis also suggested that vitamin D is ineffective to reduce risk of falls by 10%.1 After calling for patient-level metaanalyses to do subgroup analyses, and previously supporting treatment recommendations based on such analyses2 their suggestion to stop torturing these data torturing seems disingenuous. Although a patientlevel meta-analysis might generate new hypotheses to be tested in trials, such analyses for fracture3,4 added little to knowledge that was based on trial-level analyses. Trial
www.thelancet.com/diabetes-endocrinology Vol 2 July 2014
sequential analyses advance the field because they are novel and provide estimates of optimum sample size and futility, in this case indicating that additional trials of similar design are unlikely to change the finding that vitamin D supplementation does not have an important effect on fall risk1 or other important skeletal or non-skeletal events.5 These results are relevant for researchers, funders, and policymakers, and they should discourage widespread indiscriminate vitamin D supplementation and futile research endeavours, while they can also encourage a focus on better defining of vitamin D deficiency. Funded by the Health Research Council (HRC) of New Zealand. MJB is the recipient of a Sir Charles Hercus Health Research Fellowship. AG, GDG, and IRR declare no competing interests.
*Mark J Bolland, Andrew Grey, Greg D Gamble, Ian R Reid [email protected] Department of Medicine, University of Auckland, Auckland 1142, New Zealand 1
2 3
4
5
Bolland MJ, Grey A, Gamble GD, Reid IR. Vitamin D supplementation and falls: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; published online April 24. http://dx.doi.org/10.1016/S22138587(14)70068-3. Davies JS, Poole CD. Vitamin D: too much of a good thing? Br J Gen Pract 2014; 64: 8–9. DIPART (Vitamin D Individual Patient Analysis of Randomized Trials) Group. Patient level pooled analysis of 68 500 patients from seven major vitamin D fracture trials in US and Europe. BMJ 2010; 340: b5463. Bischoff-Ferrari HA, Willett WC, Orav EJ, Lips P, Meunier PJ, Lyons RA, et al. A pooled analysis of vitamin D dose requirements for fracture prevention. N Engl J Med 2012; 367: 40–49. Bolland MJ, Grey A, Gamble GD, Reid IR. The effect of vitamin D supplementation on skeletal, vascular, or cancer outcomes: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; 2: 307–20.
541
between individuals at baseline will start to have possible effects on the results. The conclusion that lifestyle interventions for people with impaired glucose tolerance lowers diabetesrelated deaths, and that these effects are true in women especially, thus seem somewhat premature. I declare no competing interests.
Mattias Brunström [email protected] Department of Public Health and Clinical Medicine, Division of Medicine, Umea University, Umea 901 87-SE, Sweden 1
2
3
4
Li G, Zhang P, Wang J, et al. Cardiovascular mortality, all-cause mortality, and diabetes incidence after lifestyle intervention for people with impaired glucose tolerance in the Da Qing Diabetes Prevention Study: a 23-year follow-up study. Lancet Diabetes Endocrinol 2014; 2: 474–80. Bassler D, Montori VM, Briel M, et al. Reflections on meta-analyses involving trials stopped early for benefit: Is there a problem and if so, what is it? Stat Methods Med Res 2013; 22: 159–68. Schulz KF, Grimes DA. Multiplicity in randomised trials II: subgroup and interim analyses. Lancet 2005; 365: 1657–61. Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle interventions to prevent diabetes in the China Da Qing Diabetes Prevention Study: a 20-year follow-up study. Lancet 2008; 371: 1783–89.
Authors’ reply We thank Mattias Bronström for his interest in our Article.1 The Da Qing Diabetes Prevention Study was not terminated prematurely, but ran the planned full course of 6 years of intervention.2 Subsequently, 20 years after the initial randomisation, we designed and implemented an observational follow-up study with prespecified outcomes and reported the results on the macrocomplications and microcomplications of diabetes.3,4 Because of potentially very important, but non-significant, trends in mortality seen in the 20 year followup, a subsequent follow-up at 23 years was specifically designed to determine whether the death rates between the lifestyle intervention and control groups continued to diverge.1 The observation of a greater effect of lifestyle intervention in women, which was the result of a post-hoc 540
analysis, was not anticipated and the explanations remain unclear. Given its level of statistical significance, we believe that this result needed to be reported. However, we do share some of Bronström’s hesitation to accept its magnitude at face value. More importantly, as the first report of a randomised controlled trial to show statistically significant reduction in mortality rates after lifestyle intervention in patients with impaired glucose tolerance,1 replication of the findings from continuing studies such as The Diabetes Prevention Program Outcomes Study (DPPOS)5 will be important to determine the extent to which our results can be generalised. We declare no competing interests.
*Guangwei Li, Ping Zhang, Edward W Gregg, Michael M Engelgau, Peter H Bennett [email protected] Department of Endocrinology, China–Japan Friendship Hospital, Beijing 100029, China (GL); Center of Endocrinology and Cardiovascular Disease, National Center of Cardiology and Fuwai Hospital, Beijing, China (GL); Division of Diabetes Translation (PZ, EWG) and Center for Global Health (MME), Centers for Disease Control and Prevention, Atlanta, GA 30333, USA; and Phoenix Epidemiology and Clinical Research Branch, National Institute of Diabetes and Digestive and Kidney Diseases, Phoenix, AZ, USA (PHB) 1
2
3
4
5
Li G, Zhang P, Wang J, et al. Cardiovascular mortality, all-cause mortality, and diabetes incidence after lifestyle intervention for people with impaired glucose tolerance in the Da Qing Diabetes Prevention Study: a 23-year follow-up study. Lancet Diabetes Endocrinol 2014; 2: 474–80. Pan XR, Li GW, Hu YH, et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: 537–44. Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle interventions to prevent diabetes in the China Da Qing Diabetes Prevention Study: a 20-year follow-up study. Lancet 2008; 371: 1783–89. Gong Q, Gregg EW, Wang J, et al. Long-term effects of a randomised trial of a 6-year lifestyle intervention in impaired glucose tolerance on diabetes-related microvascular complications: the China Da Qing Diabetes Prevention Outcome Study. Diabetologia 2011; 54: 300–07. Knowler WC, Fowler SE, Hamman RF, et al. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86.
Vitamin D and falls In their Article in The Lancet Diabetes & Endocrinology (April 24), Bolland and colleagues1 conclude that “at present, there is little justification for prescribing vitamin D supplements to prevent falls”. The authors’ key underpinning assumption rests on their belief that a threshold risk reduction of 15% “is the smallest effect that is clinically relevant for an individual”. In our view, this statement naturally warrants examination in context. We are carrying out an economic evaluation of alternative public health strategies with regards to vitamin D supplementation. A simple 5 year budget impact model using UK inputs for outcomes2 and costs, taking into account institutional care for older adults,3 suggests that even if the effect of vitamin D supplementation (800 IU every day) on falls was as little as a 6% relative risk reduction, use of this intervention in individuals older than 75 years would save costs for the UK National Health Service. In this conservative scenario, more than 6000 deaths would be avoided and a further 7000 older adults could remain living independently (with a higher quality of life). Usefully, the work of Bolland’s group in Auckland does show the substantial heterogeneity between the few trials available. Some specific subgroups might benefit, but traditional aggregate data meta-analysis is poorly placed to detect these benefits reliably. The statistical power of aggregate data meta-analysis in simulation studies4 can be dramatically and consistently lower than that in equivalent individual patient data meta-analyses, with little agreement between the estimates obtained from the two methods. Only in meta-analyses of large numbers of large trials does meta-regression detect differential treatment effects between risk groups with any consistency. Therefore, while meta-analysis of summary data might be adequate to estimate a single pooled treatment
www.thelancet.com/diabetes-endocrinology Vol 2 July 2014
Correspondence
effect or to investigate study-level characteristics, when investigators are assessing whether patient characteristics are related to treatment, individual patient data analysis will generally be necessary to discover any such relations. In this regard, we agree with the authors that widespread, indiscriminate, supplementation with vitamin D has a poor evidence base, but disagree with their arbitrary threshold of clinical relevance. We suggest that to determine which patients to target with treatment, more trials of consistent design are needed. In the meantime, the authors should heed their own advice regarding meta-analyses of vitamin D and “stop torturing these data”5. JSD declares consultancy work and speaker fees for Prostrakan, Internis Pharmaceuticals, and Sandoz.
*J Stephen Davies, Chris D Poole [email protected] University Hospital of Wales, Cardiff CF14 4XW, UK (JSD) and Cochrane Institute for Primary Care and Public Health, Cardiff University, Cardiff, UK (CDP) 1
2
3
4
5
Bolland MJ, Grey A, Gamble GD, Reid IR. Vitamin D supplementation and falls: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; published online April 24. http://dx.doi. org/10.1016/S2213-8587(14)70068-3. Scuffham P, Chaplin S LR. Incidence and costs of unintentional falls in older people in the United Kingdom. J Epidemiol Community Health 2003; 57: 740–44. Bebbington A, Darton R, Netten A. Elderly people admitted to residential and nursing homes: 42 months on. PSSRU. November, 2000. http://www.pssru.ac.uk/pdf/p051.pdf (accessed June 9, 2014). Lambert PC, Sutton AJ, Abrams KR, Jones DR. A comparison of summary patient-level covariates in meta-regression with individual patient data meta-analysis. J Clin Epidemiol 2002; 55: 86–94. Grey A, Bolland M, Reid I. Vitamin D dose requirements for fracture prevention. N Engl J Med 2012; 367: 1367–70.
Authors’ reply In their Correspondence, Davies and Poole estimate that a 6% reduction in fall risk associated with vitamin D supplementation would be cost effective. This calculation might be correct, but conflating cost effectiveness for the population with individual clinical benefit is unwise. The absolute benefit of a 6% relative risk reduction for individuals is very small. For a group with a 10% per year risk of falls, about 27 of 100 people will fall in 3 years. If the entire group took a treatment that reduced the risk by 6%, 25 people will fall by 3 years— ie, of 100 people, two benefit and 98 do not. It seems unlikely that people would take a drug regularly for 3 years for such a small benefit, only because it is regarded as cost-effective. Furthermore, proving that vitamin D reduces falls by 6% would need very large trials or meta-analyses: the optimum sample size for a trial sequential analysis that uses this threshold is roughly 57 000 people. Notably, trial sequential analysis also suggested that vitamin D is ineffective to reduce risk of falls by 10%.1 After calling for patient-level metaanalyses to do subgroup analyses, and previously supporting treatment recommendations based on such analyses2 their suggestion to stop torturing these data torturing seems disingenuous. Although a patientlevel meta-analysis might generate new hypotheses to be tested in trials, such analyses for fracture3,4 added little to knowledge that was based on trial-level analyses. Trial
www.thelancet.com/diabetes-endocrinology Vol 2 July 2014
sequential analyses advance the field because they are novel and provide estimates of optimum sample size and futility, in this case indicating that additional trials of similar design are unlikely to change the finding that vitamin D supplementation does not have an important effect on fall risk1 or other important skeletal or non-skeletal events.5 These results are relevant for researchers, funders, and policymakers, and they should discourage widespread indiscriminate vitamin D supplementation and futile research endeavours, while they can also encourage a focus on better defining of vitamin D deficiency. Funded by the Health Research Council (HRC) of New Zealand. MJB is the recipient of a Sir Charles Hercus Health Research Fellowship. AG, GDG, and IRR declare no competing interests.
*Mark J Bolland, Andrew Grey, Greg D Gamble, Ian R Reid [email protected] Department of Medicine, University of Auckland, Auckland 1142, New Zealand 1
2 3
4
5
Bolland MJ, Grey A, Gamble GD, Reid IR. Vitamin D supplementation and falls: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; published online April 24. http://dx.doi.org/10.1016/S22138587(14)70068-3. Davies JS, Poole CD. Vitamin D: too much of a good thing? Br J Gen Pract 2014; 64: 8–9. DIPART (Vitamin D Individual Patient Analysis of Randomized Trials) Group. Patient level pooled analysis of 68 500 patients from seven major vitamin D fracture trials in US and Europe. BMJ 2010; 340: b5463. Bischoff-Ferrari HA, Willett WC, Orav EJ, Lips P, Meunier PJ, Lyons RA, et al. A pooled analysis of vitamin D dose requirements for fracture prevention. N Engl J Med 2012; 367: 40–49. Bolland MJ, Grey A, Gamble GD, Reid IR. The effect of vitamin D supplementation on skeletal, vascular, or cancer outcomes: a trial sequential meta-analysis. Lancet Diabetes Endocrinol 2014; 2: 307–20.
541
