The Governor and Attorney General of South Dakota (“the State”), along with two intervening crisis pregnancy centers and two of their personnel (collectively “Intervenors”), appeal the district court’s permanent injunction barring enforcement of a South Dakota statute requiring the disclosure to patients seeking abortions of an “[ijncreased risk of suicide ideation and suicide,” see S.D.C.L. § 34-23A-10.1(l)(e)(ii) (“suicide advisory”), and the underlying grant of summary judgment in favor of Planned Parenthood of Minnesota, North Dakota, South Dakota and its medical director Dr. Carol E. Ball (collectively “Planned Parenthood”) that this advisory would unduly burden abortion rights and would violate physicians’ First Amendment right to be free from compelled speech. For the reasons discussed below, we reverse.
I.
In 2005, South Dakota enacted House Bill 1166 (“the Act”), amending the requirements for obtaining informed consent to an abortion as codified in S.D.C.L. § 34-23A-10.1. Section 7 of the Act requires physicians, in the course of obtaining informed consent, to provide certain information to the patient seeking an abortion. In June 2005, Planned Parenthood sued to prevent the Act from taking effect, contending that several of its provisions constituted an undue burden on abortion rights and facially violated patients’ and physicians’ free speech rights, while other provisions were unconstitutionally vague. After the district court preliminarily enjoined the Act and a divided panel of this court affirmed, this court sitting en banc vacated the preliminary injunction and remanded for further proceedings. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 530 F.3d 724 (8th Cir.2008) (en banc).
On remand, the parties filed cross-motions for summary judgment with respect to the challenged provisions. The district court ruled that a biological disclosure, see §§ 34-23A-10.1(l)(b), 34-23A-R4), and a medical emergency exception, see § 34-23A-10.1, were facially sound with respect to the First Amendment and imposed no undue burden, while disclosures regarding the protected relationship between the patient and the unborn child, see § 34-23A-10.1(l)(c), (d), and the suicide advisory, see § 34-23A-10.1(l)(e)(ii), failed to meet both constitutional requirements. The district court also held that a requirement to disclose “all known medical risks of the procedure,” see § 34-23A-10.1(l)(e), was not *893unconstitutionally vague, but that a requirement to disclose “statistically significant risk factors,” see id,., was.
Planned Parenthood appealed the district court’s decision on the biological disclosure and the “all known medical risks” disclosure, while the State and Intervenors appealed the district court’s decision on the relationship disclosures and the suicide advisory. A panel of this court affirmed unanimously with respect to the biological disclosure and the “all known medical risks” disclosure, reversed unanimously with respect to the relationship disclosures, and affirmed in a divided decision as to the suicide advisory. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 653 F.3d 662 (8th Cir.2011). We granted this rehearing en banc solely on the issue of the suicide advisory.1
II.
We review a grant of summary judgment de novo. Missouri ex rel. Nixon v. Am. Blast Fax, Inc., 323 F.3d 649, 653 (8th Cir.2003). In addition, we review constitutional challenges and questions of statutory interpretation de novo. McDermott v. Royal, 613 F.3d 1192, 1193 (8th Cir.2010) (per curiam).
Planned Parenthood contends that requiring a physician to present the suicide advisory imposes an undue burden on abortion rights and violates the free speech rights of the physician. “[W]hen the government requires [as part of the informed consent process] ... the giving of truthful, nonmisleading information about the nature of the procedure, the attendant health risks and those of childbirth,” and other information broadly relevant to the decision to have an abortion, it does not impose an undue burden on abortion rights, even if the disclosure “might cause the woman to choose childbirth over abortion.” Planned Parenthood of Se. Penn. v. Casey, 505 U.S. 833, 882-83, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992). Moreover, “the physician’s First Amendment rights not to speak are implicated, but only as part of the practice of medicine, subject to reasonable licensing and regulation by the State.” Id. at 884, 112 S.Ct. 2791 (citations omitted). Thus, with respect to First Amendment concerns, “while the State cannot compel an individual simply to speak the State’s ideological message, it can use its regulatory authority to require a physician to provide truthful, non-misleading information relevant to a patient’s decision to have an abortion, even if that information might also encourage the patient to choose childbirth over abortion.” Rounds, 530 F.3d at 734-35; accord Tex. Med. Providers Performing Abortion Servs. v. Lakey, 667 F.3d 570, 576-77 (5th Cir.2012).
In short, to succeed on either its undue burden or compelled speech claims, Planned Parenthood must show that the disclosure at issue “is either untruthful, misleading or not relevant to the patient’s decision to have an abortion.” Rounds, 530 F.3d at 735. To evaluate the constitutional merits of the suicide advisory, we will examine first what disclosure actually is required, second whether that disclosure is truthful, and third whether it is non-misleading and relevant to the patient’s decision to have an abortion.
III.
Section 34-23A-10.1 requires a physician seeking to perform an abortion to present to the patient:
(1) A statement in writing providing the following information:
*894(e) A description of all known medical risks of the procedure and statistically significant risk factors to which the pregnant woman would be subjected, including:
(i) Depression and related psychological distress;
(ii) Increased risk of suicide ideation and suicide;
Planned Parenthood argues, and the district court agreed, that subsection (ii) must be construed to require a disclosure of a conclusive causal link between abortion and suicide. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 650 F.Supp.2d 972, 982 (D.S.D.2009). However, no language in subsection (ii), or in the heading of section 10.1(l)(e), refers to such a causal link. “The intent of a statute is determined from what the legislature said, rather than what the courts think it should have said, and the court must confíne itself to the language used.” Langdeau v. Langdeau, 751 N.W.2d 722, 727 (S.D.2008) (quoting US W. Commc’ns, Inc. v. Pub. Utils. Comm’n, 505 N.W.2d 115, 123 (S.D.1993)).
Here, the language actually used by the legislature — “medical risks,” “statistically significant risk factors,” “[i]n-creased risk” — denotes risk in a medical context. Moreover, while the heading of subsection (e) refers broadly to “all known medical risks of the [abortion] procedure ... including” those listed in its subsections, the suicide advisory is the only subsection to further incorporate the more precise phrase “[¡Increased risk.” See § 34-23A-10.1(l)(e)(ii). Therefore, we must presume that the term “increased risk” has a more precise meaning than the umbrella term “risk” by itself. See Maynard v. Heeren, 563 N.W.2d 830, 835 (S.D.1997) (“[N]o wordage should be found to be surplus. No provision can be left without meaning. If possible, effect should be given to every part and every word.” (quoting Cummings v. Mickelson, 495 N.W.2d 493, 500 (S.D.1993))); see also FCC v. AT & T Inc., — U.S.-, 131 S.Ct. 1177, 1183, 179 L.Ed.2d 132 (2011) (recognizing that, in construing a statute, “two words together may assume a more particular meaning than those words in isolation”). The term “increased risk” is not defined in the statute, and it has more than one reasonable definition in the medical field. South Dakota law requires that such a term “must be construed according to its accepted usage, and a strained, unpractical or absurd result is to be avoided.” Peters v. Spearfish ETJ Planning Comm’n, 567 N.W.2d 880, 885 (S.D.1997).
As a result, the disclosure actually required by the suicide advisory depends upon the accepted usage of the term “increased risk” in the relevant medical field. We turn to the medical literature and expert evidence in the record to discern the accepted usage of the term “increased risk” in the applicable medical context, with an eye towards whether that accepted usage necessarily implies proof of causation.
The peer-reviewed medical literature in the record on the topic of suicide and abortion consistently uses the term “increased risk” to refer to a relatively higher probability of an adverse outcome in one group compared to other groups — that is, to “relative risk.” See Stedman’s Medical Dictionary 1701 (28th ed. 2006) (defining relative risk as “the ratio of the r[isk] of disease among those exposed to a r[isk] factor to the r[isk] among those not exposed”). For example, one study compared the rate of suicide for women who had received an induced abortion with the rates of suicide for two other groups, women who had given birth and women who had miscarried. See Ex. 60, Mika Gissler *895et al., Suicides After Pregnancy in Finland, 1987-94, 313 Brit. Med. J. 1431, 1432 (1996), ECF No. 172-3.2 That study characterized its finding of a vastly higher suicide rate for women who received an induced abortion as “an increased risk of suicide.” Id. at 1434. Another study compared the rate of, inter alia, suicide ideation in women who had received an induced abortion with the rates for women who had given birth and for women who had not become pregnant. See Ex. 61, David M. Fergusson et al., Abortion in Young Women and Subsequent Mental Health, 47 J. Child Psychol. & Psychiatry 16, 19 (2006), ECF No. 172-4. That study concluded, “Certainly in this study, those young women who had abortions appeared to be at moderately increased risk of both concurrent and subsequent mental health problems when compared with equivalent groups of pregnant or non-pregnant peers.” Id. at 23 (emphasis added).
The discussion of risk in the medical context provided by Intervenors’ expert also supports the conclusion that the term “increased risk” refers to the comparison of two groups, or relative risk:
Assessment of degree of risk is often expressed in terms of absolute risk, which relates to the chance of developing a disease over a time-period (e.g., a 10% lifetime risk of suicide) or in terms of relative risk, which is a comparison of the probability of an adverse outcome in two groups. For example, abortion would be considered an increased risk for suicide if the relative risk is significantly higher for women who abort compared to women who give birth or never have children.
Coleman Decl. ¶ 6, Jul. 6, 2006, ECF No. 189 (emphases added). Based on the “accepted usage” of the term in the relevant field, Peters, 567 N.W.2d at 885, the term “increased risk” in subsection (ii) indicates that the “relative risk” definition is the one intended by the legislature for the suicide advisory.
Noticeably absent from the contextual definition of “increased risk” is a requirement for conclusive proof of causation. This stands to reason, because, as explained by the Intervenors’ expert:
When examining complex human psychological and physical health outcomes, such as depression and suicidal behavior, identification of a single, precise causal mechanism applicable to all situations is not possible----
Given this inherent complexity, sound epidemiological evidence is nevertheless derived by identifying those variables which are most strongly linked with adverse mental or physical health outcomes for large groups of individuals.
Coleman Decl. ¶¶ 5-6, Jul. 6, 2006. While such evidence of relative risk eventually may prove direct causation as further experiments rule out plausible competing explanations, see id. at ¶ 9, conclusive proof of causation is not required in order for the identification of a medical risk.
Even the evidence upon which Planned Parenthood heavily relies is consistent with the “relative risk” definition of “increased risk,” with no requirement for proof of causation. For example, the report of the American Psychological Association’s (“APA”) Task Force on Mental Health and Abortion, Branson Decl. Ex. A, Sept. 8, 2008, ECF Nos. 283-3, 283-4 (hereinafter “APA Report”), decries the “tendency to confuse a risk and a cause” as a “logical fallacy.” APA Report at 31. As another example, Planned Parenthood submitted into the record a letter to a medical *896journal from one of the researchers mentioned above. While the researcher emphasized that his studies linking suicide and abortion did not prove causation, he resolutely reiterated his finding of “increased risk.” Mika Gissler et al., Letter to the Editor: Pregnancy-Related Violent Deaths, 27 Scand. J. Pub. Health 1:54, 55 (1999), ECF No. 206-10. It would be nonsensical for those in the field to distinguish a relationship of “increased risk” from one of causation if the term “risk” itself was equivalent to causation.
In the face of this extensive evidence of the accepted usage of the term “increased risk,” Planned Parenthood makes two arguments as to why the suicide advisory should be read to require a disclosure of proof of causation. First, it argues that the statute refers to the “increased risk of suicide ideation and suicide” as a risk “to which the pregnant woman would be subjected ” by the abortion procedure, see § 34-23A-10.1(l)(e) (emphasis added), implying that the abortion procedure directly subjects the patient to, or causes, the result. A relevant rule of statutory construction, however, holds that “a limiting clause or phrase ... should ordinarily be read as modifying only the noun or phrase that it immediately follows.” Barnhart v. Thomas, 540 U.S. 20, 26, 124 S.Ct. 376, 157 L.Ed.2d 333 (2003). Under that rule, the phrase “to which the pregnant woman would be subjected” modifies only the immediately preceding phrase “statistically significant risk factors” (which is not at issue here), not the phrase “all known medical risks of the procedure” (of which the “increased risk of suicide ideation and suicide” is a listed example). See § 34-23A-10.1(l)(e).
Moreover, even if the phrase “to which the pregnant woman would be subjected” is construed to modify the “increased risk” language, it would not advance Planned Parenthood’s argument because the result to which the pregnant woman would be subjected is the increased risk. In other words, the abortion procedure causes the patient to become a member of a group for which an increased risk is documented relative to other groups. This does not imply proof that the abortion procedure directly causes the adverse outcome in those cases where the risk materializes. There is a very real difference between (1) a statement that an action places an individual at an increased risk for an adverse outcome, and (2) a statement that, if the individual experiences the adverse outcome, the action will have been the direct cause.3
Second, Planned Parenthood relies on the “established principle of statutory construction that, where the wording of an act is changed by amendment, it is evidential of an intent that the words shall have a different construction.” Lewis & Clark Rural Water Sys., Inc. v. Seeba, 709 N.W.2d 824, 831 (S.D.2006) (quoting S.D. Subsequent Injury Fund v. Federated Mut. Ins., Inc., 605 N.W.2d 166, 170 (S.D.2000)). The informed-consent statute in *897effect prior to the Act required the disclosure of “[t]he particular medical risks associated with the particular abortion procedure to be employed including, when medically accurate, the risks of infection, hemorrhage, danger to subsequent pregnancies, and infertility.” S.D.C.L. § 34-23A-10.1(l)(b) (2004) (emphasis added). The Act expanded this subject matter into a new, four-part subsection:
(e) A description of all known medical risks of the procedure and statistically significant risk factors to which the pregnant woman would be subjected, including:
(i) Depression and related psychological distress;
(ii) Increased risk of suicide ideation and suicide;
(in) A statement setting forth an accurate rate of deaths due to abortions, including all deaths in which the abortion procedure was a substantial contributing factor;
(iv) All other known medical risks to the physical health of the woman, including the risk of infection, hemorrhage, danger to subsequent pregnancies, and infertility!)]
§ 34-23A-10.1(l)(e) (2005). Because this provision as amended by the Act no longer includes the word “associated,” Planned Parenthood asks us to conclude that the legislature intended the term “increased risk” to imply proof of causation, rather than that the procedure and the adverse outcome are merely “associated” by a correlative relationship such as relative risk.
We certainly agree that the amendments to the medical-risks provision are “evidential of an intent that the words shall have a different construction,” Lewis & Clark Rural Water Sys., 709 N.W.2d at 831 (quoting S.D. Subsequent Injury Fund, 605 N.W.2d at 170), but in this case that different construction does not hinge on the removal of one word. Instead, the Act effects essentially a complete rewriting of the former § 34-23A-10.1(l)(b) (2004), removing thirteen of the original twenty-eight words and adding seventy new words, including an entirely new introduction requiring a description of “all known medical risks” and a listing of three new specific areas of concern in subsections (i)(iii). Taken as a whole, these sweeping changes to the language of the provision express the legislature’s intent to address a much broader range of specific medical risks in the required disclosure, not to implicitly sever the term “increased risk” from its accepted usage in the medical field. See Lewis & Clark Rural Water Sys., 709 N.W.2d at 830 (“[T]he true intent of the legislature in enacting laws ... is ascertained primarily from the language employed in the statute.” (quoting Sanford v. Sanford, 694 N.W.2d 283, 287 (2005))). Indeed, where only fifteen words of original language remain in an amended provision of eighty-five words, ascribing such an effect to the removal of a single word would go far beyond any use of the cited rule of statutory construction of which we are aware. See, e.g., S.D. Subsequent Injury Fund, 605 N.W.2d at 170-71 (applying the statutory-amendment rule of construction to a seventy-one-word statute to which four new words were added and two were changed).
Finally, even if the language of the suicide advisory also reasonably could be construed to require a disclosure of a causal link, we would be faced with “varying constructions of the South Dakota statute, ‘by [one] of which grave and doubtful constitutional questions arise and by [the other] of which such questions are avoided.’” Rounds, 653 F.3d at 669 (quoting United States v. Adler, 590 F.3d 581, 583 (8th Cir.2009)). In such a situation, our “duty is to adopt the latter,” id. (quoting Adler, 590 F.3d at 583), and “[t]his is *898especially so since ‘[i]n evaluating a facial challenge to a state law, a federal court must ... consider any limiting construction that a, state ... enforcement agency has proffered,’ ” id. (quoting Kolender v. Lawson, 461 U.S. 352, 355, 103 S.Ct. 1855, 75 L.Ed.2d 903 (1983)). As a result, we would be called to apply the “relative risk” construction of increased risk over a construction that required disclosure of a causal link.
To summarize, in subsection (ii), the legislature expressly required the disclosure of an “increased risk,” not a causal link. Based on the accepted usage of the term “increased risk” in the relevant medical field, the usage of that term in the context of § 34 — 23A—10.1(l)(e)(ii) does not imply a disclosure of a causal relationship. Instead, subsection (ii) requires a disclosure simply that the risk of suicide and suicide ideation is higher among women who abort compared to women in other relevant groups, such as women who give birth or do not become pregnant.
IV.
With regard to whether the required disclosure is truthful, see Rounds, 530 F.3d at 735, the State submitted into the record numerous studies published in peer-reviewed medical journals that demonstrate a statistically significant correlation between abortion and suicide. The studies were published in respected, peer-reviewed journals such as the Obstetrical and Gynecological Survey, the British Medical Journal, the Journal of Child Psychology and Psychiatry, the Southern Medical Journal, and the European Journal of Public Health, and there is no indication that the peer-review process was compromised for the studies at issue. See Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 593-94, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) (“The fact of publication (or lack thereof) in a peer reviewed journal ... [is] a relevant, though not dis-positive, consideration in assessing the scientific validity of a particular technique or methodology on which an opinion is premised” because “submission to the scrutiny of the scientific community ... increases the likelihood that substantive flaws in methodology will be detected.”).
Planned Parenthood argues that these studies do not examine the correlation between abortion and suicide in sufficient detail to prove a causal link (as discussed in more detail in Part V), but, as we concluded above, the suicide advisory does not require disclosure of a causal link. With regard to the accuracy of the correlation itself, there is nothing in the record to suggest that the underlying data or calculations in any of these studies are flawed. For example, Planned Parenthood’s own expert, Dr. Nada Stotland, admitted that one of the studies, which determined a suicide rate after abortion of 31.9 per 100,-000 as compared to a suicide rate after live birth of 5.0 per 100,000, “indicates an association; not causation, but an association” between abortion and suicide. Stotland Dep. 283:22-284:9, ECF No. 152-12.4 When asked if she had “any quarrel with the validity of that association,” Dr. Stotland replied that she did not. Id. at 284:11-13.
Based on the record, the studies submitted by the State are sufficiently reliable to support the truth of the proposition that the relative risk of suicide and suicide ideation is higher for women who abort *899their pregnancies compared to women who give birth or have not become pregnant. It also is worth noting that Planned Parenthood does not challenge the disclosure that “[depression and related psychological distress” is a “known medical risk[ ] of the [abortion] procedure.” S.D.C.L. § 34-23A-10.1(l)(e)(i); see also Gonzales v. Carhart, 550 U.S. 124, 159, 127 S.Ct. 1610, 167 L.Ed.2d 480 (2007) (noting that “[s]evere depression and loss of esteem can follow” an abortion). As a matter of common sense, the onset of depression and psychological distress also would increase one’s risk of suicide and suicide ideation. See, e.g., Ottar Bjerkeset et al., Gender Differences in the Association of Mixed Anxiety and Depression with Suicide, 192 Brit. J. Psychiatry 474, 474 (2008) (“Depression is thought to be the most important antecedent of suicide.... ”). Thus, there appears to be little dispute about the truthfulness of the required disclosure.
Finally, Planned Parenthood contends that the suicide advisory is not truthful because an increased risk of suicide after abortion is not “known” as required by the statute. See S.D.C.L. § 34-23A-10.1(l)(e) (requiring disclosure of “[a]ll known medical risks of the procedure”); Rounds, 653 F.3d at 673 (“ ‘[K]nown’ means generally recognized, proved, or familiar to all.”). Once again, however, this contention is premised on Planned Parenthood’s argument that the term “increased risk” implies a causal link that is not generally “known.” Because the statute does not require the disclosure of any causal link, Planned Parenthood’s argument on this point is misdirected. The record indicates that the disclosure actually required — that the relative risk of suicide and suicide ideation is higher for women who abort compared to women in other relevant groups — is generally “known.” For example, the ninety-one-page APA Report, on which Planned Parenthood relies extensively, was commissioned for the sole purpose of analyzing that “known” risk in more detail. See APA Report at 5.
As a result, we hold that the disclosure facially mandated by the suicide advisory is truthful.
V.
Despite the extensive evidence in the record of an “increased risk” of suicide, Planned Parenthood contends that disclosure of the increased risk would be misleading and irrelevant to a patient seeking an abortion, see Rounds, 530 F.3d at 735, because some authorities have indicated that there is no direct causal link. In particular, Planned Parenthood argues that it is more plausible that certain underlying factors, such as pre-existing mental health problems, predispose some women both to have unwanted pregnancies and to have suicidal tendencies, resulting in a misleading correlation between abortion and suicide that has no direct causal component. Under this view, the required disclosure would be misleading or irrelevant to the decision to have an abortion because the patient’s decision would not alter the underlying factors that actually cause the observed increased risk of suicide.
As an initial matter, the standard medical practice, as reflected in the record, is to recognize a strongly correlated adverse outcome as a “risk” while further studies are conducted to clarify whether various underlying factors play causal roles. See, e.g., Coleman Decl. ¶¶ 9-11, Jul. 6, 2006. In contravention of that standard practice, Planned Parenthood argues that the mere existence of underlying factors proscribes the disclosure of suicide as a risk related to abortion. However, there is no constitutional requirement to invert the traditional understanding of “risk” by requiring, where abortion is involved, that conclusive understanding of causation be obtained first. Indeed, the Supreme *900Court “has given state and federal legislatures wide discretion to pass legislation in areas where there is medical and scientific uncertainty,” and “[m]edieal uncertainty does not foreclose the exercise of legislative power in the abortion context any more than it does in other contexts.” Gonzales, 550 U.S. at 163-64, 127 S.Ct. 1610. In particular, “a requirement that a doctor give a woman certain information as part of obtaining her consent to an abortion is, for constitutional purposes, no different from a requirement that a doctor give certain specific information about any medical procedure.” Casey, 505 U.S. at 884, 112 S.Ct. 2791. There is no basis in the “non-misleading” and “relevant” requirements of Casey for imposing a new, stricter definition of medical risk — a standard that requires certainty of causation— simply because the medical procedure at issue is abortion.
Thus, the truthful disclosure regarding increased risk cannot be unconstitutionally misleading or irrelevant simply because of some degree of “medical and scientific uncertainty,” Gonzales, 550 U.S. at 163, 127 S.Ct. 1610, as to whether abortion plays a causal role in the observed correlation between abortion and suicide. Instead, Planned Parenthood would have to show that any “medical and scientific uncertainty” has been resolved into a certainty against a causal role for abortion. In other words, in order to render the suicide advisory unconstitutionally misleading or irrelevant, Planned Parenthood would have to show that abortion has been ruled out, to a degree of scientifically accepted certainty, as a statistically significant causal factor in post-abortion suicides. An examination of Planned Parenthood’s evidence reveals that it has not met this burden.
First, Planned Parenthood points out that the label approved by the Food and Drug Administration (“FDA”) for the abortion-inducing drug Mifeprex (mifepristone, also known as RU-486) does not list suicide or suicide ideation as a risk of using the drug, despite FDA labeling regulations requiring the listing of, inter alia, all “clinically significant adverse reactions” and “other potential safety hazards.” See 21 C.F.R. § 201.57(c)(6)(i). However, an FDA-approved label does not represent the definitive or exclusive list of risks associated with a drug. The record before us does not show whether any evidence of the link between abortion and suicide was submitted to the FDA, nor does it provide details of the FDA’s analysis, if any, of the link. Thus, the FDA-approved label for Mifeprex yields no information as to whether abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.
Second, Planned Parenthood argues, and the district court found, that the American College of Obstetricians and Gynecologists (“ACOG”), a well-known professional medical organization, “rejects any suggestion that increased risk of suicide and suicide ideation are known risks of abortion.” See Rounds, 650 F.Supp.2d at 983. Unfortunately, there was no evidence from ACOG in the record for the district court to consider. The only evidence in the record pertaining to ACOG’s position is a second-hand reference in a 2005 report by the State’s expert, Dr. Elizabeth M. Shadigian, that quoted two sentences from a single ACOG Practice Bulletin: “Long-term risks sometimes attributed to surgical abortion include potential effects on ... psychological sequelae. However, the medical literature, when carefully evaluated, clearly demonstrates no significant negative impact on any of these factors with surgical abortion.” Elizabeth M. Shadigian, Report to the S.D. Task Force to Study Abortion 4, Sept. 21, 2005, ECF No. 177-4 (hereinafter “Shadigian Report”); see also Ex. O, Shadigian Dep. 137-38, ECF No. 147-15 (quoting the recitation of those lines in the Shadigian Re*901port). Dr. Shadigian further reported her opinion that ACOG’s statement was erroneous and that “ACOG seems to claim that they have adequately evaluated the medical literature, but they do not consider our study or the many other studies we evaluated.” Shadigian Report at 5. There is no other evidence in the record as to what “medical literature” ACOG considered, in what fashion it was “carefully evaluated,” whether suicide was one of the “psychological sequelae” considered, whether ACOG’s analysis received any independent peer review, or indeed whether a “Practice Bulletin” purports to be grounded in any sort of reliable scientific method at all. The two unsupported sentences from an ACOG Practice Bulletin lend no credence to the argument that abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.
Third, Planned Parenthood cites the previously mentioned APA Report. The six-person Task Force on Mental Health and Abortion that authored the APA Report reviewed “50 papers published in peer-reviewed journals between 1990 and 2007 that analyzed empirical data of a quantitative nature on psychological experiences associated with induced abortion, compared to an alternative.” APA Report at 64. For some of the studies that found increased mental health risks associated with abortion, the APA Report identifies perceived methodological deficiencies, including an inability to limit the comparison group to women who carried unplanned or unwanted pregnancies to term. See id. at 68. Based on one study that attempted to account for that variable, the report states that “the best scientific evidence indicates that the relative risk of mental health problems among adult women who have an unplanned pregnancy is no greater if they have an elective first-trimester abortion than if they deliver that pregnancy.” Id. (emphases in original). In the very same sentence, however, the report states that the published literature could not provide “unequivocal evidence regarding the relative mental health risks associated with abortion per se compared to its alternatives (childbirth of an unplanned pregnancy).” Id.
The State and Intervenors argue that the APA Report is deficient in several respects. While the APA Report alleges methodological flaws in all of the studies that found a strong link between abortion and adverse mental health outcomes, it does not systematically list or analyze those flaws for each study considered. Instead, the report uses a handful of studies as illustrative examples. The State and Intervenors contend that this lack of rigor allowed the APA Report to analyze studies that found abortion to be “a benign experience for most women” less stringently than studies that found abortion to cause adverse effects. Coleman Deck ¶ 14, Sept. 16, 2008, ECF No. 290-3. For example, while the APA Report suggests that the studies showing increased risk did not compare women receiving abortions to women who carried unplanned pregnancies to term, at least three studies purportedly considered by the task force did use such a control group, and each of those studies still “definitively indicated that abortion was associated with more mental health problems.” Id. at ¶ 19. The APA Report also does not acknowledge that some of the studies showing increased risk did statistically control for other potential causal factors such as history of depression, anxiety, suicide ideation, childhood sexual abuse, physical abuse, child neuroticism, and low self-esteem. Id. at ¶ 15(c).5 As another example, although a high rate of attrition (i.e., the loss of subjects from a *902long-term study before the study is complete) is typically regarded as a methodological weakness, the APA Report downplays the significance of attrition, possibly because “the studies with the highest attrition rates ... are also the ones that provide little evidence of negative effects” of abortion. Id. at ¶ 15(d). A number of published authors in the field contacted the APA to point out these problems and ask that the APA Report be retracted. Id. at ¶¶ 28-29.
At a minimum, it appears that many published authors in the field do not accept the opinion of the APA’s six-person task force that the “best evidence” suggests that there is no real significance to the link between abortion and suicide. Even if one accepts the findings in the APA Report at face value, however, the crux of the matter is that while the APA Report states that the evidence available at the time of its review is not “sufficient to support the claim that an observed association between abortion history and mental health was caused by the abortion,” id. at 6 (emphasis added), it also concludes that the published literature is inconclusive and more research is needed “to disentangle confounding factors and establish relative risks of abortion compared to its alternatives,” id. at 72; see also id. at 68 (admitting that the published literature could not provide “unequivocal evidence regarding the relative mental health risks associated with abortion per se compared to its alternatives (childbirth of an unplanned pregnancy)”). In other words, while the APA Report finds that studies to date have not established with certainty that abortion is a causal factor in post-abortion suicide, it also acknowledges that abortion has not been ruled out as a causal factor and that currently available studies are inadequate for that purpose.6 Thus, the APA Report provides no support for the proposition that abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.
Finally, the dissent relies on six recent publications submitted to this Court by Planned Parenthood as a supplement to the district court record. While the dissent suggests that these more recent publications have eliminated any uncertainty about the causal role of abortion in the increased risk of suicide, post at 909-10, the publications add little of value to the record. As an initial matter, three of the publications7 are not new analyses of data, *903but rather reviews or surveys of existing studies. As with the APA Report, it is difficult to identify a solid objective basis for the criteria employed in these reviews to identify the “best” studies and discount the others, and in the fine print they sometimes remain equivocal about the role of abortion as a causal factor. See, e.g., Robinson, supra, at 277 (“For women who have more significant [psychological] problems, the causal contribution of the abortion is not clear; a wide range of factors, both internal and external, affect women’s responses — and interact in complex ways.”).
The three remaining supplemental publications actually provide new analysis, but each suffers from apparent weaknesses. One of the publications, Julia R. Steinberg et al., Does the Outcome of a First Pregnancy Predict Depression, Suicidal Ideation, or Lower Self-Esteem? Data from the National Comorbidity Survey, 81 Am. J. Orthopsychiatry 193 (2011) (“Steinberg I”), compared self-reported mental health problems for women who carried their first pregnancy to term with women who aborted their first pregnancy, see id. at 194, while attempting to control for prepregnancy mental health, experience of sexual violence, and age at first pregnancy, see id. at 197. Mental health problems were classified as pre- or post-delivery or abortion of the first pregnancy. Id. at 195. Thus, if a woman delivered a first pregnancy, aborted a subsequent pregnancy, and suffered an adverse mental health outcome after the abortion, her adverse outcome nevertheless was classified by Steinberg I as belonging to the “delivery” comparison group, rather than the “abortion” comparison group. According to data from the Guttmacher Institute, however, approximately 45 to 47 percent of women obtaining their first abortion have previously carried at least one pregnancy to term. See Jones et al., Repeat Abortion in the United States, Guttmacher Institute, 18 (Nov. 2006), http://www. guttmacher.org/pubs/2006/ll/21/or29.pdf (listing number of prior births for women having a first abortion, based on two sets of data collected by the Guttmacher Institute at abortion provider locations). As a result, Steinberg I almost certainly shifts the outcomes for a significant number of women who aborted their second or subsequent pregnancies from the “abortion” comparison group to the “delivery” comparison group, rendering its comparison of mental health outcomes unreliable.
Another study, Julia R. Steinberg & Lawrence B. Finer, Examining the Association of Abortion History and Current Mental Health: A Reanalysis of the National Comorbidity Survey Using a Common-Riskr-Factors Model, 72 Soc. Sci. & Med. 72 (2011) (“Steinberg II”), compared self-reported mental health problems for women who had been pregnant but never aborted with those women who had aborted one or more pregnancies, while attempting to control for pre-pregnancy mental health, experience of violence, and age at first pregnancy. See id. at 77. However, Steinberg II identified post-abortion or post-delivery mental health outcomes only as those problems the participants reported they were experiencing at the time of the survey, rather than considering mental health problems that participants reported as occurring at any time after an abortion or delivery. See id. at 76-77. Therefore, it appears that Stein-berg II addresses only an arbitrarily limit*904ed window of the women’s mental health histories.
Finally, Trine Munk-Olsen et ah, Induced First-Trimester Abortion and Risk of Mental Disorder; 364 New Eng. J. Med. 332 (2011), extracted data from the Danish Civil Registration System and the Danish Psychiatric Central Register. The study considered data only for women who had no history of inpatient treatment for mental illness in their lifetime prior to the nine-month period preceding either a first abortion or a first delivery of a pregnancy. See id. at 334. For that group of women, the study analyzed data regarding inpatient or outpatient mental health contacts during the nine-month period preceding abortion or delivery and the twelve-month period following abortion or delivery. See id. While the study observed a significantly higher number of psychiatric visits after abortion as compared to after delivery, see id. at 335, it “found no significant increase in the incidence rate of psychiatric contact in the 12 months after an induced first-trimester abortion as compared with the 9-month period before the abortion” and concluded that the higher incidence in post-abortive women compared to post-delivery women likely was due solely to higher pre-existing levels of psychiatric problems for the women who sought abortions, see id. at 336. This conclusion apparently begs the question, however, by assuming that any mental distress occurring in the nine-month period prior to an abortion procedure was completely unrelated to the abortion. It seems just as plausible to assume, particularly in a population selected for having no adverse mental health history prior to that time period, that for at least some of the women, psychological distress in that time period arose in part because they had decided, or already were facing pressure from others, to undergo the abortion. Under this interpretation, the study actually tends to confirm the legislative finding that women who seek abortions are “often under stress and pressures from circumstances and from other persons, and that there exists a need for special protection of the rights of such pregnant women.” S.D.C.L. § 34-23A-1.5.
We acknowledge that these studies, like the studies relied upon by the State and Intervenors, have strengths as well as weaknesses. Like all studies on the topic, they must make use of imperfect data that typically was collected for entirely different purposes, and they must attempt to glean some insight through the application of sophisticated statistical techniques and informed assumptions. While the studies all agree that the relative risk of suicide is higher among women who abort compared to women who give birth or do not become pregnant, they diverge as to the extent to which other underlying factors account for that link. We express no opinion as to whether some of the studies are more reliable than others; instead, we hold only that the state legislature, rather than a federal court, is in the best position to weigh the divergent results and come to a conclusion about the best way to protect its populace. So long as the means chosen by the state does not impose an unconstitutional burden on women seeking abortions or their physicians, we have no basis to interfere.
In summary, although the record reflects “medical and scientific uncertainty,” Gonzales, 550 U.S. at 163, 127 S.Ct. 1610, as to whether abortion itself is a causal factor in the observed correlation between abortion and suicide, there is nothing in the record to suggest that abortion as a cause per se has been ruled out with certainty. As a result, the disclosure of the observed correlation as an “increased risk” is not unconstitutionally misleading or irrelevant under Casey and Gonzales. Indeed, physicians who provide abortions *905should be capable of reviewing the research in the field, understanding the difference between relative risk and proof of causation, and explaining it correctly to their patients. Cf. Rounds, 530 F.3d at 736 (holding that the subject matter of the biological disclosure “should be clear in context to a physician”).8 In the end, “[t]he point of informed consent laws is to allow the patient to evaluate her condition and render her best decision under difficult circumstances. Denying her up to date medical information is more of an abuse to her ability to decide than providing the information.” Lakey, 667 F.3d at 579.
Accordingly, we hold that the suicide advisory is non-misleading and relevant to the patient’s decision to have an abortion.
VI.
In conclusion, we hold that the requirements of S.D.C.L. § 34 — 23A—10.1 (1)(e)(ii) are satisfied by a disclosure that the relative risk of suicide and suicide ideation is higher for women who abort compared to women in other relevant groups, as described in the relevant medical research. The statute does not require the physician to disclose that a causal link between abortion and suicide has been proved. The disclosure is truthful, as evidenced by a multitude of studies published in peer-reviewed medical journals that found an increased risk of suicide for women who had received abortions compared to women who gave birth, miscarried, or never became pregnant. Various studies found this correlation to hold even when controlling for the effects of other potential causal factors for suicide, including pre-existing depression, anxiety, suicide ideation, childhood sexual abuse, physical abuse, child neuroticism, and low self-esteem.
Moreover, the suicide advisory is non-misleading and relevant to the patient’s decision to have an abortion, as required by Casey. It is a typical medical practice to inform patients of statistically significant risks that have been associated with a procedure through medical research, even if causation has not been proved definitively.9 While Planned Parenthood points to uncertainty as to whether abortion itself is a causal factor in the observed correlation to suicide, as opposed to other underlying factors that tend to be associated independently with both abortion and suicide, the Supreme Court “has given state and federal legislatures wide *906discretion to pass legislation in areas where there is medical and scientific uncertainty,” including “in the abortion context.” Gonzales, 550 U.S. at 163-64, 127 S.Ct. 1610. Thus, a truthful disclosure cannot be unconstitutionally misleading or irrelevant simply because some degree of medical and scientific uncertainty persists. To be sure, informed consent requirements “must be calculated to inform [a] woman’s free choice, not hinder it,” Casey, 505 U.S. at 877, 112 S.Ct. 2791, but there is no unconstitutional hindrance of the woman’s choice where, as here, the State merely is using “its regulatory authority to require a physician to provide truthful, non-misleading information relevant to a patient’s decision to have an abortion, even if that information might also encourage the patient to choose childbirth over abortion,” Rounds, 530 F.3d at 735.
On its face, the suicide advisory presents neither an undue burden on abortion rights nor a violation of physicians’ free speech rights. Accordingly, we reverse the district court’s grant of summary judgment to Planned Parenthood with respect to S.D.C.L. § 34-23A-10.1 (l)(e)(ii), direct the entry of summary judgment for the State as to that provision, and vacate the permanent injunction against the enforcement of that provision.
. Apart from Section II.C of the panel opinion, which addresses the suicide advisory and was vacated by our order taking this matter en banc, the panel opinion remains in force.
. All cited exhibit numbers and ECF designations refer to the summary judgment exhibit numbers and ECF document heading numbers, respectively, in the district court record, No. 05-cv-4077 (D.S.D.).
. This difference may be better illustrated by an example less contentious than abortion. One recent study found that prolonged television viewing resulted in an “increased risk” of mortality for individuals in any given age group. See Anders Grontved et al., Television Viewing and Risk of Type 2 Diabetes, Cardiovascular Disease, and All-Cause Mortality, 305 J. Am. Med. Assoc. 23:2448 (2011). We would not demand proof that television viewing itself directly caused the adverse outcome (for example, proof of an actual decline in the health of heart muscle tissue to a fatal level during viewing) before acknowledging that a prolonged television viewer is “subjected” to the increased risk of mortality. Indeed, a measure of increased risk based on a discrete, easily reportable event such as television viewing is useful precisely because of the difficulty of tracing exactly whether and how a given action combines with other factors to directly “cause” a particular death.
. With regard to another potential comparison group, the cited study also determined a suicide rate among women of reproductive age who did not become pregnant as in the range of 11.8 to 13.3 per 100,000. See Mika Gissler et al.. Injury Deaths, Suicides and Homicides Associated with Pregnancy, Finland 1987-2000, 15 Eur. J. Pub. Health 5:459, 460 (2005), ECF No. 147-18.
. The dissent notes that one study authored by Coleman and cited in her declaration on this issue later was found to contain errors. Post at 910. However, Coleman’s declaration *902cites various studies by other authors that control for these other potential causal factors and nevertheless find a persistent link between abortion and increased mental health problems. See Coleman Decl. ¶¶ 22-24, Sept. 16, 2008, ECF No. 290-3. Her declaration was not rebutted with respect to those studies.
. While the APA awaits methodologically perfect research on the effect of “unwanted” or "unplanned” pregnancies, others have suggested that such perfection may not be achievable, because "pregnancies that are aborted frequently were initially intended by one or both partners and pregnancies that are initially unintended often become wanted as the pregnancy progresses, rendering assessment of wantedness/intentedness [sic] subject to considerable change over time.” Coleman Decl. ¶ 15, Jul. 6, 2006. In addition, "pregnancy wantedness/intendedness is open to multiple subjective interpretations.” Id. at ¶ 16. The APA Report does not specify what sort of data on these variables would be acceptable to resolve the issue to the APA's satisfaction, and the report even seems to conflate the entirely separate concepts of whether a pregnancy is "wanted” with whether it was initially "planned” or "intended.” See, e.g., APA Report at 64 ("These studies were evaluated with respect to their ability to draw sound conclusions about the relative mental health risks associated with abortion compared to alternative courses of action that can be pursued by a woman facing a similar circumstance (e.g., an unwanted or unintended pregnancy).”).
. National Collaborating Centre for Mental Health, Induced Abortion and Mental Health: A Systemic Review of the Mental Health Out*903comes of Induced Abortion, Including Their Prevalence and Associated Factors (2011); Royal College of Obstetricians and Gynaecologists, The Care of Women Requesting Induced Abortion (2011); Gail Erlick Robinson et al., Is There an "Abortion Trauma Syndrome?" Critiquing the Evidence, 17 Harv. Rev. Psychiatry 268 (2009).
. To the extent the dissent suggests that a patient will receive a physician’s detailed explanation of the disclosure only if she seeks additional explanation and clarification, see post at 911-12, we disagree. The statute requires the physician to provide, in writing, "[a] description” of the risks at issue, § 34-23A — 10.1 (l)(e), not just a recitation of the statutory language. Contrary to the dissent’s reference to a "judicial attempt to direct the content of the conversation between a patient and her doctor,” post at 912, we recognize that the legislature left the precise content of that description to the physician's discretion.
. We disagree with the dissent's suggestion that this is a new standard or theory about the nature of an informed consent advisory. See post at 911. Instead, statements about “increased risk” in the absence of conclusive proof of causation have been treated as material in a variety of contexts. See, e.g., Brock v. Merrell Dow Pharms., Inc., 874 F.2d 307, 312 (explaining that if studies establish, within an acceptable confidence interval, that those who use a pharmaceutical have a relative risk of greater than 1.0 — that is, an increased risk — of an adverse outcome, those studies might be considered sufficient to support a jury verdict of liability on a failure-to-warn claim), modified on reh’g, 884 F.2d 166 (5th Cir.1989); 21 C.F.R. § 201.80(e) (requiring that prescription drug "labeling shall be revised to include a warning as soon as there is reasonable evidence of an association of a serious hazard with a drug; a causal relationship need not have been proved”). The decision of the South Dakota legislature that the increased risk at issue here likewise merits an advisory is not atypical.