In all subgroups, hazard ratios for intensive treatment using traditional Cox model were similar to the original SPRINT report. Using the cJM approach, intensive treatment decreased the risk for the primary outcome among all subgroups at the start of follow-up. However, the effect lost its significance after 1.3 and 3.4 years among participants with and without baseline CKD, after 1.1 and 3.5 years among women and men, after 1.8 and 3.1 years among black and nonblack individuals, after 2.1 and 3.4 years among individuals less than 75 years and at least 75 years, after 1.3 and 3.4 years among participants with and without prevalent CVD, after 2.5, 2.0, and 1.8 years for individuals with baseline SBP of 132 mmHg or less, between 133 and 144 mmHg, and at least 145 mmHg, respectively (Fig. 6a–l). Appendix Table S1, http://links.lww.com/HJH/B41 presents the hazard ratios (95% CIs) at the start and end of follow-up for all subgroups.
SAEs occurred in 96.3% (n = 516) of participants who suffered the primary outcome. Appendix Table S2, http://links.lww.com/HJH/B41, shows distribution of SAEs by subgroup. Using Cox proportional hazards analyses (hazard ratio; 95% CI), hypotension (1.71; 1.26, 2.33), electrolyte abnormalities (1.38; 1.07, 1.79), and acute renal failure (1.68; 1.33, 2.12) were significantly associated with intensive treatment (Table 2). In the cJM, the interaction term for having experienced SAEs during follow-up in the overall population was significant (P for interaction SAEs × treatment <0.0001). Therefore, we stratified the cJM analyses based on occurrence of SAEs during follow-up. Cox analyses hazard ratios (95% CI) for intensive treatment in groups with and without SAEs were 0.74 (0.62, 0.88) and 0.25 (0.084, 0.75), respectively. Using the cJM approach, the hazard ratios (95% CI) at the start of follow-up were 0.60 (0.50, 0.72) and 0.19 (0.06, 0.63) for the groups with and without SAEs, respectively (Table S1, http://links.lww.com/HJH/B41). The effect lost significance after 3.4 years for participants with SAEs but remained significant until 4.2 years of follow-up for the non-SAEs group. The wider 95% CI for the non-SAEs group reflects the small number of primary outcomes in this group (n = 20; 4% of all primary outcomes) (Fig. 6m and n). Finally, we evaluated the differential effect of SAEs in the SPRINT primary outcome among interventions groups using Cox proportional analysis. We found significantly three times larger effects of SAEs on the hazard ratio for primary outcome, in the intensive treatment group (hazard ratio: 96.95; P < 0.000) compared with standard treatment (hazard ratio: 33.42; P < 0.000) in the overall SPRINT population.
Our secondary analysis of SPRINT confirmed that intensive hypertension treatment lowered the risk for the primary outcome at start of follow-up. However, the initial beneficial effect was lost during follow-up in the overall population and particularly among participants with prevalent CKD or CVD, women, black individuals, younger participants, and those with SBP above 132 mmHg at baseline. The beneficial effect of intensive treatment was also lost earlier among patients who suffered SAEs during follow-up.
Conventionally, trials correlate the baseline BP values with outcomes of interest. The original SPRINT analysis showed a 25% reduction in the primary outcome for intensive treatment, using the traditional Cox approach assuming that the benefits remain constant over time. However, besides higher BP at baseline, cumulative exposure to BP and its variability are important risk factors for CVD and kidney dysfunction [4–6]. Our analyses simultaneously took into account the dependency and association between repeated SBP measurements and time-to-event and allowed for evaluation of both direct and indirect (i.e. through SBP) effects of the intensive treatment [14–17]. When cumulative effect of SBP and its variability, both within individuals and between treatment groups, was taken into account, the initial beneficial effect of intensive treatment was lost during follow-up. Importantly, recent secondary analyses of ONTARGET and TRASCEND trials showed a higher predictive value for a composite mean SBP over time compared with baseline or event-preceding or time-updated SBP , which substantiates our approach. Based on experimental studies, high BP variability induces a chronic inflammatory state through activation of the myocardial angiotensin-converting enzyme, increasing the expression of monocyte-protein-1 and transforming growth factor-B, resulting in ventricular hypertrophy, remodeling and dysfunction, perivascular fibrosis, endothelial injury, and kidney dysfunction [19–23].
Three recent studies investigating the association of visit-to-visit BP variability with primary SPRINT outcome and adverse events have led to conflicting results. Chang et al. showed no association between BP variability with primary SPRINT outcome but a significant association with all-cause mortality. This study, however, included only the SBP measurements between 3 and 18 months of follow-up and discarded about 42% (n = 238) of the primary SPRINT outcomes. Moreover, they adjusted for multiple covariates disregarding the previous treatment randomization. Goyal et al. showed SBP variability to be independently associated with higher risk of hyponatremia among SPRINT participants. In another post-hoc analysis among a subset of SPRINT participants with baseline CKD, DBP variability was associated with the primary outcome and with major SAEs .
The beneficial effect of intensive treatment was lost earlier among specific subgroups in our analyses; including CKD participants, women, and individuals of black ethnicity. Previous studies have observed larger SBP variability among these groups, linking it to a higher vascular risk among these individuals [23,27]. Their larger SBP variability might explain earlier loss of beneficial effect of intensive treatment among these individuals. Compared with older participants, individuals younger than 75 years lost the beneficial effect of intensive treatment earlier. Although SBP variability increases with age, younger individuals have shown a greater susceptibility to target organ damage resulting from SBP variability . Moreover, older patients might respond better to medications such as diuretics due to their beneficial impact on outcomes such as congestive heart failure which is one component of the primary SPRINT outcome . These factors might explain earlier loss of beneficial impact of intensive treatment among younger individuals in our study. We also observed that individuals with SBP more than 132 mmHg at baseline and during follow-up lost the beneficial impact of intensive treatment earlier compared with those with SBP of 132 or less. This could be attributed to a higher SBP variability among individuals with SBP more than 132 mmHg due to larger fluctuations in the number or dose of prescribed antihypertensive medications in this group. It is important to mention that the 95% CI for participants with prevalent CKD or CVD, women, black individuals was substantially wider than the comparison groups. However, the slope of the graphs for participants with prevalent CKD or CVD, women, black individuals were clearly larger compared with non-CKD, non-CVD, male, nonblack race subgroups, respectively (P interaction <0.0001) (Fig. 6, Supplemental Fig. 6, http://links.lww.com/HJH/B41).
Intensive BP lowering could lead to adverse events altering the efficacy of this intervention during follow-up. Our study showed less benefit for intensive treatment among individuals who experienced SAEs during follow-up (Fig. 6m and n). Although the proportion of participants who suffered SAEs was similar between the intensive and standard treatment groups, type of adverse event was different. More severe adverse events including hypotension, electrolyte abnormalities, and acute kidney injury occurred more often in the intensive treatment group. In addition to cumulative SBP and its variability, development of SAEs could partly explain loss of initial beneficial effect for intensive treatment over time. A secondary analysis of SPRINT among participants with normal renal function at baseline showed a 1.2 ratio for developing CKD per preventing one cardiovascular event . The risk for mortality and CVD among patients with renal dysfunction is between 1.2–1.8 and 1.9–2.9, respectively . Projecting the SPRINT eligibility criteria to the 1999–2006 National Health and Nutrition Examination Survey showed that intensive treatment prevents 107 500 deaths per-year but increases the number of patients with SAEs to 222 600 per-year . Notably, SAEs occurred in the majority of SPRINT participants who suffered the primary outcome (96.3%). In the intensive treatment group, SAEs were associated with SPRINT primary outcome three times more than the SAEs in the standard group. If SAEs increase the risk of primary outcome, the harms of intensive hypertension treatment might offset its potential benefits.
New guidelines for management of BP, redefine the therapeutic target as BP less than 130/80 mmHg [31,32]. For primary prevention, the guidelines recommend pharmacology treatment among individuals with BP more than 130/80 mmHg and cardiovascular risk more than 10% or those with cardiovascular risk less than 10% but BP more than 140/90 mmHg. In secondary prevention settings, pharmacology treatment is recommended for BP more than 130/80. However, our results in the subgroup of SPRINT participants with CVD history showed earlier loss of beneficial impact of intensive SBP treatment over time than for non-CVD participants.
Despite the observed increasing tendency in the hazard ratios over time, as the SPRINT terminated after median 3.26 years of follow-up (range 0–4.5 years), our findings are only applicable to this time-window. 96.3% of patients who developed a primary outcome suffered SAEs during follow-up. This led to small number of events and limited power for the analyses among participants without SAEs.
Concerns have been raised that the BP measurements in SPRINT might not be directly comparable with those of other trials and not readily applicable to clinical settings. The measurement of BP in the SPRINT was unattended at the majority of study sites . Assessment of 24-h ambulatory BP monitoring in a subset of SPRINT participants, demonstrated that daytime ambulatory SBP was higher than clinic SBP, the agreement between daytime ambulatory SBP and clinic SBP was poor, and the difference in ambulatory SBP between the two SPRINT treatment groups was lower than the difference measured by clinic SBP . Although a subsequent analyses of the SPRINT reported that the SPRINT results were insensitive to whether or not BP measurements were made in an attended manner , it has been suggested that treatment arms in SPRINT could translate into clinic SBP of 132 versus 144 mmHg [36,37].
Major strength of our study is the use of a robust statistical model which allows us to maintain the initial SPRINT randomization in our analyses. In addition, our approach allows for evaluation of the cumulative impact of SBP and its variability (both intraindividual and between groups) as well as SAEs on the primary SPRINT outcome, taking into account that hazard ratios may change over time . An additional benefit of using joint model analysis is that postrandomization BP measurements are treated as an outcome (and not as a covariate), the joint likelihood of the BP measurements and the time to the primary endpoint are also completely specified, thus providing valid estimates of the treatment effect. During development of the statistical model and for construction of different SBP trajectories over time, we specifically took into account the initial decrease in SBP at the beginning of follow-up. As model specification and goodness of fit are fundamental for the validity of our results, supplemental statistical material details all the steps we followed for development of our statistical models.
Finally, we are aware that our results may be considered controversial, however, they are in line with what was originally published by the SPRINT group (Fig. 4 original publication in The New England Journal of Medicine) , which showed that intensive treatment did not significantly reduce cardiovascular risk in patients with CKD, younger participants, women, black individuals, CVD, and those with baseline SBP more than 132 mmHg. These groups are the same ones in which we have found that the protective benefit of intensive treatment is lost early during follow-up. Thus, loss of beneficial effect occurred earlier in those who did not significantly benefit from intensive treatment in the original SPRINT. These findings further increase confidence in the validity of our results.
In conclusion, intensive SBP treatment lowered the risk for the primary SPRINT outcome at the start of follow-up. However, the initial beneficial effect was lost during follow-up in the overall population and particularly among participants with prevalent CKD or CVD, women, black individuals, younger participants, those with baseline SBP more than 132 mmHg, and patients who suffered SAEs.
Our results call for caution regarding universal recommendations for intensive BP treatment, particularly among specific subgroups. In addition to potential adverse effects from intensive treatment, the impact of cumulative SBP as well as intraindividual SBP variability should not be dismissed. As the tenet of medicine ‘Primum non nocere’ must prevail, longer term clinical trials, with focus on sustained beneficial effects of intensive interventions over time and on patient safety are needed. cJM analysis is a novel and not frequently considered approach for assessment of clinical trial data. This method adds a time-varying perspective which approaches the conditions encountered in daily clinical practice.
To main researchers of SPRINT, data sharing initiative from The New England Journal of Medicine and BioLINCC (NHLBI) for allow us to do this secondary analysis. Clinical trial registration: URL: https://www.clinicaltrials.gov. Unique identifier: NCT01206062. O.L.R.-O. receives a scholarship from Colciencias – Colombia and support from Trustfond Erasmus University, the Netherlands and Universidad Industrial de Santander UIS, Colombia.
Conflicts of interest
There are no conflicts of interest.
1. Kannel WB, Dawber TR, Kagan A, Revotskie N, Stokes J 3rd. Factors of risk in the development of coronary heart disease – six year follow-up experience. The Framingham Study. Ann Intern Med
2. [No authors listed]. Multiple risk factor intervention trial. Risk factor changes and mortality results. Multiple Risk Factor Intervention Trial Research Group. JAMA
3. Stevens SL, Wood S, Koshiaris C, Law K, Glasziou P, Stevens RJ, McManus RJ. Blood pressure variability and cardiovascular disease: systematic review and meta-analysis. BMJ
4. Gosmanova EO, Mikkelsen MK, Molnar MZ, Lu JL, Yessayan LT, Kalantar-Zadeh K, Kovesdy CP. Association of systolic blood pressure variability with mortality, coronary heart disease, stroke, and renal disease. J Am Coll Cardiol
5. Li W, Jin C, Vaidya A, Wu Y, Rexrode K, Zheng X, et al. Blood pressure trajectories and the risk of intracerebral hemorrhage and cerebral infarction: a prospective study. Hypertension
6. Howard SC, Rothwell PM. Reproducibility of measures of visit-to-visit variability in blood pressure after transient ischaemic attack or minor stroke. Cerebrovasc Dis
7. Muntner P, Joyce C, Levitan EB, Holt E, Shimbo D, Webber LS, et al. Reproducibility of visit-to-visit variability of blood pressure measured as part of routine clinical care. J Hypertens
8. Pfeffer MA, McMurray JJ. Lessons in uncertainty and humility – clinical trials involving hypertension. N Engl J Med
9. Weber MA, Poulter NR, Schutte AE, Burrell LM, Horiuchi M, Prabhakaran D, et al. Is it time to reappraise blood pressure thresholds and targets? A statement from the International Society of Hypertension – a global perspective. Hypertension
10. Bundy JD, Li C, Stuchlik P, Bu X, Kelly TN, Mills KT, et al. Systolic blood pressure reduction and risk of cardiovascular disease and mortality: a systematic review and network meta-analysis. JAMA Cardiol
11. Lewington S, Clarke R, Qizilbash N, Peto R, Collins R. Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet
12. Bress AP, Kramer H, Khatib R, Beddhu S, Cheung AK, Hess R, et al. Potential deaths averted and serious adverse events incurred from adoption of the SPRINT (Systolic Blood Pressure Intervention Trial) intensive blood pressure regimen in the United States: projections from NHANES (National Health and Nutrition Examination Survey). Circulation
13. Wright JT Jr, Williamson JD, Whelton PK, Snyder JK, Sink KM, Rocco MV, et al. SPRINT Research Group. A randomized trial of intensive versus standard blood-pressure control. N Engl J Med
14. Chapman and Hall/CRC Biostatistics Series, Rizopoulos D. Joint models for longitudinal and time-to-event data: with applications in R. 1st ed.2012.
15. Rizopoulos D. JM: an R package for the joint modelling of longitudinal and time-to-event data. J Stat Softw
16. Ibrahim J, Chu H, Chen L. Basic concepts and methods for joint models of longitudinal and survival data. J Clin Oncol
17. Tsiatis A, Davidian M. Joint modeling of longitudinal and time-to-event data: an overview. Statistica Sinica
18. Bohm M, Schumacher H, Teo KK, Lonn EM, Mahfoud F, Mann JFE, et al. Achieved blood pressure and cardiovascular outcomes in high-risk patients: results from ONTARGET and TRANSCEND trials. Lancet
19. Kudo H, Kai H, Kajimoto H, Koga M, Takayama N, Mori T, et al. Exaggerated blood pressure variability superimposed on hypertension aggravates cardiac remodeling in rats via angiotensin II system-mediated chronic inflammation. Hypertension
20. Yasuoka S, Kai H, Kajimoto H, Kudo H, Takayama N, Anegawa T, et al. Blood pressure variability activates cardiac mineralocorticoid receptor and induces cardiac remodeling in hypertensive rats. Circ J
21. Aoki Y, Kai H, Kajimoto H, Kudo H, Takayama N, Yasuoka S, et al. Large blood pressure variability aggravates arteriolosclerosis and cortical sclerotic changes in the kidney in hypertensive rats. Circ J
22. Hodgson JM, Woodman RJ, Croft KD, Ward NC, Bondonno CP, Puddey IB, et al. Relationships of vascular function with measures of ambulatory blood pressure variation. Atherosclerosis
23. Diaz KM, Veerabhadrappa P, Kashem MA, Thakkar SR, Feairheller DL, Sturgeon KM, et al. Visit-to-visit and 24-h blood pressure variability: association with endothelial and smooth muscle function in African Americans. J Hum Hypertens
24. Chang TI, Reboussin DM, Chertow GM, Cheung AK, Cushman WC, Kostis WJ, et al. Visit-to-visit office blood pressure variability and cardiovascular outcomes in SPRINT (Systolic Blood Pressure Intervention Trial). Hypertension
25. Goyal A, Mezue K, Rangaswami J. Visit-to-visit systolic blood pressure variability predicts treatment-related adverse event of hyponatremia in SPRINT. Cardiovasc Ther
26. Mezue K, Goyal A, Pressman GS, Horrow JC, Rangaswami J. Blood pressure variability predicts adverse events and cardiovascular outcomes in chronic kidney disease: a post-hoc analysis of the SPRINT trial. Am J Hypertens
27. Rothwell PM, Howard SC, Dolan E, O’Brien E, Dobson JE, Dahlof B, et al. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet
28. Bulpitt CJ, Beckett NS, Peters R, Leonetti G, Gergova V, Fagard R, et al. Blood pressure control in the Hypertension in the Very Elderly Trial (HYVET). J Hum Hypertens
29. Beddhu S, Rocco MV, Toto R, Craven TE, Greene T, Bhatt U, et al. Effects of intensive systolic blood pressure control on kidney and cardiovascular outcomes in persons without kidney disease: a secondary analysis of a randomized trial. Ann Intern Med
30. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med
31. Whelton PK, Carey RM, Aronow WS, Casey DE Jr, Collins KJ, Dennison Himmelfarb C, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol
32. Bakris G, Sorrentino M. Redefining hypertension – assessing the new blood-pressure guidelines. N Engl J Med
33. Kjeldsen SE, Lund-Johansen P, Nilsson PM, Mancia G. Unattended blood pressure measurements in the systolic blood pressure intervention trial. Implications for entry and achieved blood pressure values compared with other trials. Hypertension
34. Drawz PE, Pajewski NM, Bates JT, Bello NA, Cushman WC, Dwyer JP, et al. Effect of intensive versus standard clinic-based hypertension management on ambulatory blood pressure. Results from the SPRINT (Systolic Blood Pressure Intervention Trial) ambulatory blood pressure study. Hypertension
35. Johnson KC, Whelton PK, Cushman WC, Cutler JA, Evans GW, Snyder JK, et al. Blood pressure measurement in SPRINT (Systolic Blood Pressure Intervention Trial). Hypertension
36. Kjeldsen SE, Mariampillai JE, Nilsson PM. Optimal blood pressure target in diabetic and nondiabetic hypertensive patients. Circ Res
37. Oparil S, Cushman WC, Johnson KC, Kitzman DW, Whelton PK, Wright JT. Sprinting toward the optimal blood pressure target for hypertensive patients. Circ Res
38. Hernan MA. The hazards of hazard ratios. Epidemiology
* Dimitris Rizopoulos and Maryam Kavousi contributed equally to the article.
adverse effects; cumulative joint model; intensive treatment; randomized controlled trial; SBP; treatment efficacy
Supplemental Digital Content
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.