18thSeptember 2018
Authors’ response to Dr. Swapnil Hiremath’s critical remarks on our recently updated Cochrane systematic review
Luis Carlos Saiz1, Javier Gorricho2, Javier Garjón3, Mª Concepción Celaya3, Juan Erviti1, Leire Leache1
1 Unit of Innovation and Organization, Navarre Health Service, Pamplona, Spain.
2 Planning, Evaluation and Management Service, General Directorate of Health, Government of Navarre, Pamplona, Spain.
3 Drug Prescribing Service, Navarre Health Service, Pamplona, Spain.
On July 20, 2018, the first update of our Cochrane systematic review entitled ‘Blood pressure targets for the treatment of people with hypertension and cardiovascular disease’ was published [1]. A complete version of the content can be accessed here. Shortly after the publication of the review, Dr. Hiremath published serious concerns about our work on his personal Twitter account. As he is a member of the task force for Hypertension Canada, in charge of producing the Canadian Hypertension guidelines on a yearly basis, Dr. Hiremath’s opinions and arguments were read by us with great interest. In order to keep alive a constructive debate on this relevant issue, we would like to make public some brief reflections.
In our view, the twitter comments miss the core point: If you have cardiovascular disease, aiming for lower targets for blood pressure doesn’t appear to make sense.
Is a systematic review on blood pressure (BP) targets for patients with cardiovascular disease pertinent? Does it make any sense to compare <135 versus <140-160 mmHg systolic blood pressure (SBP) targets?
https://twitter.com/hswapnil/status/1021194852556464128
Cardiovascular secondary prevention can be well defined and deserves specific consideration due to its high prevalence. Thus, in our opinion the clinical question of which blood pressure targets are more appropriate in hypertensive patients with cardiovascular disease (secondary prevention) remains highly pertinent and should not be mixed with other clinical conditions. People with established cardiovascular disease are at a particular very high riskof suffering a new cardiovascular event compared to patients with other risk factors [2].
We focus on secondary prevention patients because they’re the ones most likely to benefit from alterations to their blood pressure. In fact, a J-curve effect is still being discussed for this population [3],[4] and cardiovascular risk charts usually exclude this specific population from their scores because of its differential status. Regarding the SPRINT pre-specified entry criteria [5], the subgroup presenting with clinical cardiovascular disease must be seen as clearly different from those based on the Framingham risk score, elderly people or chronic kidney disease. For instance, in this latter case, SPRINT inclusion criteria were mainly focused on people with moderate renal impairment (eGFR 30-60 ml/min/1.73m2), who were at lower cardiovascular risk (high risk instead of very high risk)according to the European guidelines [2]).
Both interventions in our review were chosen in order to be coherent with other previous Cochrane reviews also dealing with BP targets [6,7]. The rationale behind this decision gives priority to a practical perspective, by comparing the widely agreed standard practice of achieving SBP targets around 140-160 mmHg to the newer and less established intensive approaches (any target under 135 mmHg). Moreover, though the theoretical margin between 135 and 140 mmHg can be seen as too narrow, targets of studies included presented a wider range, namely 120/130 mmHg and 140/150 mmHg in the lower and higher BP target arms, respectively. This makes it possible to find meaningful differences between both interventions in our review.
In the published research it is totally feasible to establish two SBP groups: those in the lower (120/130 mmHg) and those in the higher (140/150 mmHg). Thus, we consider the review design as an appropriate approach.
Do results change when key eligibility criteria of the review are modified?
https://twitter.com/hswapnil/status/1020808358842941440
https://twitter.com/hswapnil/status/1021161740627656705
As clearly stated and justified in the review (Discussion, page 20), we decided to use pooled data of systolic, diastolic and mean targets as the main analysis but also testing whether results remained consistent when considered separately. All these analyses were performed and fully reported in the first draft review, though to achieve simplicity only combined results were finally approved for publication by the editorial board. That said, re-analysis excluding HOT and/or AASK trials can be easily performed with the published data. On page 21, it can be read that “subsequent analyses separating trials by systolic, diastolic or mean blood pressure targets did not change these results” [1]. Therefore, even excluding HOT and/or AASK trials from the analysis, no net benefit related to the intensive intervention is observed.
We consider there is nothing wrong with pooling these data and we found that even excluding two trials with diastolic and mean targets doesn’t change the results.
https://twitter.com/hswapnil/status/1021192616778522624
Also, even if only SPRINT and ACCORD trials were considered suitable to be combined, no statistically significant differences between both interventions could be found in the main outcomes of the review: overall mortality [0.95, 95%CI (0.75-1.20)], cardiovascular mortality [0.76, 95%CI (0.53-1.11)], serious adverse events [1.00, 95%CI (0.92-1.08)] or total cardiovascular events [0.84, 95%CI (0.71-1.00)]. Combining these two relevant trials found no differences between the interventions in terms of mortality or serious adverse events.
Should Hypertension Canada 2018 guidelines categorize cardiovascular disease as ‘inconclusive evidence’?
https://twitter.com/hswapnil/status/1020808400748212225
Yes. We agree with Dr. Hiremath (and Hypertension Canada guideline 2018) that there is inconclusive evidence to promote intensive BP targets for patients with prior stroke or diabetes mellitus. However we are persuaded that, according to the best direct evidence showed in our review, it is also the case for patients with coronary heart disease. The SPRINT trial is about high risk patients rather than secondary cardiovascular prevention (represented by only 17% of total participants). Thus, it is highly inadequate to assume the SPRINT results as definitive evidence for lower BP targets in this specific population. At present, there is no clinical trial on secondary prevention which results on BP targets are available.
We found that, like patients who have had a stroke or diabetes, those with established cardiovascular disease did not do any better on intensive BP control when high quality evidence is considered.
Is it a good practice to combine data from six subgroups with differences in their baseline characteristics?
https://twitter.com/hswapnil/status/1021161786379153408
Our review was only able to get reliable information from 6 subgroups, which should be the first reason to be extremely prudent before issuing a recommendation for lower targets. We have accounted for this limitation more than once, e.g. in the ‘Potential biases in the review process’ section (page 22). On the other hand, meta-analyses include studies with different characteristics very often, not only concerning baseline data but also in the way some outcomes are measured, geographical aspects, etc. In this regard, our review is no different at all from most well-performed systematic reviews.
In order to assess the main outcomes in our protocol, it is necessary to look at those specific subgroups. After doing so, no clinical differences were found.
Additionally, we find several strong points supporting our decisions on the review to bear in mind: 1) All six trials met the requested cardiovascular risk criterion (patients with established cardiovascular disease) and also fitted well the range of interventions approved by the protocol; 2) All 9484 participants (100%) complied with the pre-planned requirements thanks to the use of individual patient data for the whole review; 3) Subgroups from ACCORD, HOT, SPRINT and SPS3 trials (95% of total participants) included a large number of individuals each that helps preserve balance between groups after randomization ; 4) No significant heterogeneity among study subgroups was detected for the main estimated outcomes; and 5) Several subgroup and sensitivity analyses were performed to confirm the robustness of the results. Coming back to the SPRINT trial, results obtained from adequate subgroups are, in our opinion, more reliable than those from a trial where patients profile does not meet the review’s inclusion criteria.
Is there any other currently available systematic review that is more trustworthy?
https://twitter.com/hswapnil/status/1021161749574062080
https://twitter.com/hswapnil/status/1021161774605787136
https://twitter.com/hswapnil/status/1021162801300058112
Sadly those systematic reviews have important flaws to take into account.
We strongly disagree with Dr. Hiremath. These systematic reviews [8,9], which are also mentioned by Hypertension Canada as a key support for lower BP targets in people with cardiovascular disease, may be interesting for hypothesis generation. However, bigger is not always better when it comes to decision-making. First, they share with our report all limitations inherent to a systematic review, including Dr. Hiremath’s claims on population variability and baseline comorbidities. Second, heterogeneity among included studies was extremely high, as a result of pooling trials comparing the effects of different BP targets with trials comparing the effects of different drugs (including both head-to-head and placebo-controlled trials). Third, not only ‘immiscible’ interventions were pooled, but also strategies accepted within each intervention arm were so lax that <120 mmHg (SPRINT) and <150 mmHg (UKPDS) co-existed within the same intensive target [8]. Finally and no less important, no individual patient data were used, increasing the risk of misclassification bias. In our view, it is inappropriate to rely on this kind of data to advocate for change in clinical practice [10].
These other systematic reviews are interesting but also have very important limitations: when people of different risks and intervention characteristics are pooled in order to make a bigger meta-analysis, the obtained results may not be reliable.
Are systematic reviews better than well-conducted trials?
https://twitter.com/hswapnil/status/1021162739593404416
https://twitter.com/hswapnil/status/1021162743343149056
A well-designed, well-powered and well-conducted trial for a particular clinical question is undoubtedly a good start, though its results would always need to be confirmed at least in a second trial. Certainly a systematic review can be seen as redundant in some occasions, when the evidence from adequate clinical trials is very solid. Unfortunately this is not the case. We still need good clinical trials addressing this question [https://twitter.com/Badves/status/1021365809409503234], good systematic reviews combining the right trials, and also good guidelines issuing sound recommendations based on the best available evidence.
At present, there are two ongoing clinical trials assessing BP targets in hypertensive patients with cerebrovascular disease (ESH-CHL-SHOT [11] and RESPECT [12]). No similar projects have been identified in patients with coronary disease.
Basically, if the right studies are not performed, we’re never really going to know the full impact of the intervention.
Should the SPRINT trial be rated as ‘unclear’ in the allocation concealment domain?
https://twitter.com/hswapnil/status/1020808414773907462
No, it is an error that has already been corrected. Originally we rated SPRINT as ‘unclear’ risk of bias in the ‘random sequence generation’ domain. The reason was that (see our review, page 40) no detailed information was provided, neither in the article nor in the protocol, on the specific randomization system used in the trial (permuted blocks, minimization, other methods, etc.). Once we have access to the key information, SPRINT will be rated as ‘low’ risk of bias in this domain. We thank Dr. Hiremath for having raised this issue.
An in-depth assessment of the SPRINT trial goes far beyond the aim of this open letter. We acknowledge SPRINT is a valuable and timely NHI initiative that deserves close attention. [5]. For instance, we are delighted about the way SPRINT has estimated and reported serious adverse events (SAE). Nevertheless, some unanswered questions also need to be clarified. Two examples: SPRINT achieved an impressive SBP difference in favor of the lower target (≈15 mmHg) by taking only one drug more than the higher target (2.8 vs 1.8 antihypertensive drugs). A feasible explanation for this fact should be proposed, taking into account the difficulty to get similar results in a real clinical setting. Also, the lower target group used more chlorthalidone and less betablockers than the higher target group, which might have affected to some extent the final results. There are some issues of the SPRINT trial that still have to be assessed.
If anything, SPRINT reveals that claimed benefits in mortality of a lower blood pressure target in high-risk nondiabetic people do not lead to a decrease in total serious adverse events, which is also a critical outcome.
Final comments
When hypertensive patients with established cardiovascular disease are considered, our systematic review has found that at present, no evidence of a net health benefit can be identified between standard or lower BP targets. Importantly, we also found very limited evidence on adverse events reported by the included trials, making it even more important to have a conservative approach. Well-designed trials are needed to clarify this debate. Our systematic review has strengths and limitations, many of them already discussed in the published paper. We keep open to constructive external criticism in order to improve our document in the future as much as possible.
Bottom line: A good systematic review looks at similar interventions in similar type of patients. When we have done that, in people with established cardiovascular disease there seems to be no benefit of intensively lowering blood pressure. This has huge clinical implications in terms of overdiagnosis and overtreatment, not to mention the iatrogenic psychological harm of telling people they have ‘high’ blood pressure when it isn’t a risk factor for their health. Furthemore, when some of the included trials don’t provide reliable adverse event information, this should make anyone cautious and act accordingly.
Bibliography
1. Saiz LC, Gorricho J, Garjón J, Celaya MC, Erviti J, Leache L. Blood pressure targets for the treatment of people with hypertension and cardiovascular disease. Cochrane Database of Systematic Reviews 2018, Issue 7. Art. No.: CD010315. DOI: 10.1002/14651858.CD010315.pub3
2. Williams B, Mancia G, Spiering W, Rosei EA, Azizi M, Burnier M, et al. 2018 ESC/ESH Guidelines for the management of arterial hypertension. Eur Heart J 2018;39:3021-104. DOI: 10.1093/eurheartj/ehy339.
3. Mancia G, Grassi G. Aggressive blood pressure lowering is dangerous: the J-curve: pro-side of the argument. Hypertension 2014;63(1):29-36. [PUBMED: 24336629].
4. Verdecchia P, Angeli F, Mazzotta G, Garofoli M, Reboldi G. Aggressive blood lowering is dangerous: the J-curve: con side of the argument. Hypertension 2014;63(1):37-40. [PUBMED: 24336630].
5. The SPRINT Research Group. A randomized trial of intensive versus standard blood-pressure control. NEJM 2015. DOI: 10.1056.NEJMoa1511939.
6. Arguedas JA, Pérez MI, Wright JM. Treatment blood pressure targets for hypertension. Cochrane Database of Systematic Reviews 2009, Issue 3. Art. No.: CD004349. DOI: 10.1002/14651858.CD004349.pub2
7. Arguedas JA, Leiva V, Wright JM. Blood pressure targets for hypertension in people with diabetes mellitus. Cochrane Database of Systematic Reviews 2013, Issue 10. Art. No.: CD008277. DOI: 10.1002/14651858.CD008277.pub2
8. Ettehad D, Emdin CA, Kiran A, Anderson SG, Callender T, Emberson J, et al. Blood pressure lowering for prevention of cardiovascular disease and death: a systematic review and meta-analysis. Lancet 2016;387(10022):957-67.
9. Xie X, Atkins E, Lv J, Bennett A, Neal B, Ninomiya T, et al. Effects of intensive blood pressure lowering on cardiovascular and renal outcomes: updated systematic review and meta-analysis. Lancet 2016;387(10017):435-43.
10. Nerenberg K, Zarnke KB, Leung AA, Dasgupta K, Butalia S, McBrien K, et al. Hypertension Canada’s 2018 Guidelines for Diagnosis, Risk Assessment, Prevention, and Treatment of Hypertension in Adults and Children. Canadian Journal of Cardiology 2018;34:506-25. DOI:10.1016/j.cjca.2018.02.022
11. Zanchetti A, Liu L, Mancia G, Parati G, Grassi G, Stramba-Badiale M, et al. Blood pressure and LDL-cholesterol targets for prevention of recurrent strokes and cognitive decline in the hypertensive patient: design of the European Society of Hypertension - Chinese Hypertension League - Stroke in Hypertension Optimal Treatment randomized trial. Journal of Hypertension 2014;32(9):1888-97. DOI: 10.1097/HJH.0000000000000254; PUBMED:24979303.
12. NCT01198496. Recurrent stroke prevention clinical outcome study (RESPECTS) [Phase IV study for effect of intensive blood-pressure control using anti-hypertensive agents in essential hypertension with history of stroke]. Clinicaltrials.gov/show/NCT01198496.