We (1) adopted an a priori model consistent with a general recommendation for the use of Hartung–Knapp–Sidik–Jonkman (HKSJ) adjustment with the Paule–Mandel (PM) estimator of between-study variance (τ²) when the number of trials (k) was 10 or fewer (2, 3). However, Kivelä, Mayer, and Cornell and Mulrow highlighted that using standard HKSJ adjustment with the PM method can underestimate the uncertainty when k < 5 and τ² = 0, and in our article led to narrower CIs in evaluating the effect of reduced salt intake on all-cause mortality in normotensive patients (1–3). Considering the difficulty in generating robust estimates in such a scenario, Kivelä, Mayer, and Cornell and Mulrow encouraged us to conduct additional analyses using modified HKSJ, Bayesian, or profile likelihood methods where relevant (2).

Accordingly, we performed sensitivity analyses for all interventions and outcomes, where k > = 2 but < 5 and τ² = 0 using modified HKSJ with the PM method. Statistical analyses were conducted using “meta” commands from Stata, version 16. Consistent with the already-published statistical plan (1), statistical significance was set at 0.05 and effect sizes were reported as risk ratios (RRs) with 95% CIs. We used I² statistics to estimate the extent of unexplained heterogeneity; I² greater than 50% was considered a high degree of between-study heterogeneity.

Twenty-nine estimates were reanalyzed. As expected, CIs widened with adjustment. Two estimates changed from protective to nonsignificant: 1) low-salt diet on all-cause mortality (RR, 0.90 [95% CI, 0.34–2.36]; low certainty of evidence) in hypertensive patients and 2) cardiovascular mortality (RR, 0.67 [CI, 0.28–1.64]; low certainty of evidence) in normotensive patients. There was no important change in conclusions or certainty in evidence for the other 27 estimates (Table).

These sensitivity analyses further validate the lack of cardiovascular effects of various nutritional supplements and dietary interventions. The editorialists highlighted the controversial finding of better mortality outcomes with a low-salt diet in view of limited evidence (4). We also discussed the inconsistent data supporting the cardiovascular benefits of a low-salt diet (1). Therefore, in view of these sensitivity analyses, the certainty of evidence regarding cardiovascular effects of reduced salt intake should be downgraded. Essentially, our updated findings support the 2019 National Academies consensus study regarding insufficient evidence to establish estimates average requirement or recommended dietary allowance for sodium (5). Meta-analyses with a small number of included studies remain a challenge. It is a difficult decision to choose between the a priori selected model and a new post hoc data-driven model. In this case, the new model likely yields the more conservative and appropriate results.

我们（1）通过具有利用哈同-纳普-Sidik-Jonkman（HKSJ）的调整与研究间方差的PAULE-曼德尔（PM）估计器的一般性建议相一致的先验模型（τ ²）时，数的试验（k）为10或更少（2，3）。然而，Kivelä，迈耶和康奈尔大学和Mulrow强调，使用标准HKSJ调整与PM方法可低估不确定性时ķ  <5和τ ² = 0，并且在我们的文章中，降低血压摄入对正常血压患者全因死亡率的影响评估中的CI较窄（1-3）。考虑到在这种情况下难以生成可靠的估计，Kivelä，Mayer，Cornell和Mulrow鼓励我们在相关时使用经过修改的HKSJ，贝叶斯或概貌似然方法进行其他分析（2）。

因此，我们进行所有的干预和结果，其中敏感性分析ķ  > = 2，但<5和τ ² 使用改性HKSJ与PM方法= 0。使用来自Stata，版本16的“ meta”命令进行统计分析。与已经发布的统计计划（1）一致，统计显着性设置为0.05，且效应大小报告为具有95％CI的风险比（RRs）。我们使用I ²统计量来估计无法解释的异质性的程度；I ²大于50％被认为是研究之间的高度异质性。

重新分析了29个估计值。正如预期的那样，配置项随着调整而扩大。两种估计值从保护性改变为非显着性：1）低盐饮食对高血压患者的全因死亡率（RR，0.90 [95％CI，0.34–2.36]；证据的确定性低）和2）心血管疾病死亡率（RR，0.67） [CI，0.28–1.64]；证据不足的患者）。对于其他27个估计值，结论或证据的确定性没有重大变化（表）。

这些敏感性分析进一步证实了各种营养补充剂和饮食干预措施对心血管没有作用。鉴于证据有限，社论者强调了有争议的发现，即低盐饮食可带来更好的死亡率结果（4）。我们还讨论了不一致的数据，这些数据支持低盐饮食对心血管的益处（1）。因此，鉴于这些敏感性分析，有关降低盐摄入量的心血管作用的证据的确定性应降低。从本质上讲，我们的最新发现支持2019年美国国家科学院的共识研究，该研究缺乏足够的证据来确定估计的平均钠摄入量或建议的钠饮食建议（5）。包含少量纳入研究的荟萃分析仍然是一个挑战。在先验选择模型和新的事后数据驱动模型之间进行选择是一个艰难的决定。在这种情况下，新模型可能会产生更为保守和适当的结果。