HF732: A bill for an act prohibiting and requiring certain actions relating to abortion involving the detection of a fetal heartbeat, and including effective date provisions.
In May of 2018, I proudly signed the fetal heartbeat bill into law. Days later, it was challenged by abortion providers who do little to help women in crisis understand their choices and grossly misrepresent the service they provide as health care.
Just a few weeks ago, the Iowa Democratic party said Roe should be the floor, not the ceiling. And yet elected Democratic officials and candidates in Iowa and around the country are never pressed to answer the tough question.
Even with the fetal heartbeat bill, the fight continues for a robust culture of life that welcomes children, supports mothers, and involves fathers. As a pro-life governor, I will continue to promote policies designed to surround every person involved in a pregnancy with protection, love, and support.
But today, the most glaring injustice of all is about to be put right. Everyone understands that a heartbeat signifies life. And we understand that when it falls silent, something precious has been lost.
It is now illegal in Tennessee to abort an unborn child who has a heartbeat. Specifically, the Heartbeat Bill protects unborn children at 6 weeks gestational age who have a heartbeat, and it also protects unborn children at 8 weeks gestational age or older. These provisions in the Heartbeat Bill have an affirmative defense for medical emergencies.
The term "fetal heartbeat," as used in the anti-abortion law in Texas, is misleading and not based on science, say physicians who specialize in reproductive health. What the ultrasound machine detects in an embryo at six weeks of pregnancy is actually just electrical activity from cells that aren't yet a heart. And the sound that you "hear" is actually manufactured by the ultrasound machine. Scott Olson/Getty Images hide caption
The Texas abortion law that went into effect last fall reads: "A physician may not knowingly perform or induce an abortion on a pregnant woman if the physician detected a fetal heartbeat for the unborn child."
The law defines "fetal heartbeat" as "cardiac activity or the steady and repetitive rhythmic contraction of the fetal heart within the gestational sac" and claims that a pregnant woman could use that signal to determine "the likelihood of her unborn child surviving to full-term birth."
"When I use a stethoscope to listen to an [adult] patient's heart, the sound that I'm hearing is caused by the opening and closing of the cardiac valves," says Dr. Nisha Verma, an OB-GYN who specializes in abortion care and works at the American College of Obstetricians and Gynecologists.
"At six weeks of gestation, those valves don't exist," she explains. "The flickering that we're seeing on the ultrasound that early in the development of the pregnancy is actually electrical activity, and the sound that you 'hear' is actually manufactured by the ultrasound machine."
"What we're really detecting is a grouping of cells that are initiating some electrical activity," she explains. "In no way is this detecting a functional cardiovascular system or a functional heart."
"This is a term that is not widely used in medicine," Kern says. "I think this is an example of where we are sometimes trying to translate medical lingo in a way that patients can understand, and this is a really unfortunate side effect of this type of translation."
In fact, "fetus" isn't technically accurate at six weeks of gestation either, says Kerns, since "embryo" is the scientific term for that stage of development. Obstetricians don't usually start using the term "fetus" until at least eight weeks into the pregnancy.
Later in a pregnancy is when a clinician might use the term "fetal heartbeat," after the sound of the heart valves can be heard, she says. That sound "usually can't be heard with our Doppler machines until about 10 weeks."
The term "fetal heartbeat" has been used in laws restricting access to abortion for years. According to the Guttmacher Institute, which tracks reproductive health policy, the first such law was passed in North Dakota in 2013, but it was struck down in the courts. Since then, over a dozen states have passed similar laws, but Texas' is the first to go into effect.
The text of the Texas law claims that "fetal heartbeat has become a key medical predictor that an unborn child will reach live birth" and continues, "the pregnant woman has a compelling interest in knowing the likelihood of her unborn child surviving to full-term birth based on the presence of cardiac activity."
But obstetricians say that's not how this information is used by health care providers. "We don't use it to date a pregnancy," says Dr. Samantha Kaplan, an OB-GYN at Boston Medical Center and assistant professor at Boston University's School of Medicine.
In reality, it would be really hard for a woman to know she's pregnant before the point at which cardiac activity would be detectable by an ultrasound. She would have to be tracking her periods carefully, have regular periods, notice her period was late and then be able to quickly get an appointment with her doctor to confirm a pregnancy.
Pregnancies are dated from the start of a woman's last period. A few weeks after that, if fertilization and implantation occur, Kern says, there's "a window of a few days, maybe a week or two at the most, where you can actually detect an intrauterine pregnancy [with an ultrasound] before you detect any kind of cardiac motion or electrical activity," says Kerns.
The way pregnancies are clinically dated is not intuitive and causes people to misunderstand the real impact of these laws, says Verma. "I actually did research around the general public's understanding of 'six weeks,' " she says. "A lot of people thought that 'six weeks' referred to six weeks from your first missed period, but it's actually six weeks from your last period."
Many people do not find out they're pregnant that quickly. The women who are highly attuned to their cycle timing and able to get into a clinic promptly tend to be more affluent and educated, says Kaplan.
"Somebody who is working to get through the day and to get food on the table is not going to do that." Getting a pregnancy test, she says, "is going to be the last thing on their list, and by the time they are doing that, it's going to be too late."
Network can fail in many ways, sometimes pretty subtle(e.g. high ratio packet loss). Disrupted TCP connections takea moderately long time (about 11 minutes with defaultconfiguration on Linux, for example) to be detected by theoperating system. AMQP 0-9-1 offers a heartbeat featureto ensure that the application layer promptly finds out aboutdisrupted connections (and also completely unresponsivepeers). Heartbeats also defend against certain networkequipment which may terminate "idle" TCP connections whenthere is no activity on them for a certain period of time.
TCP keepalives is a TCP stack feature that serves a similarpurpose and can be very useful (possibly in combination with heartbeats)but requires kernel tuning in order to be practical with most operatingsystems and distributions.
The heartbeat timeout value defines after what period of timethe peer TCP connection should be considered unreachable (down) by RabbitMQand client libraries. This value is negotiated between theclient and RabbitMQ server at the time of connection. Theclient must be configured to request heartbeats.
The negotiation process works like this: the server will suggestits configurable value and the client will reconcile it with its configured value,and send the result value back. The value is in seconds,and default value suggested by RabbitMQ is 60.
A zero value indicates that a peer suggests disabling heartbeats entirely.To disable heartbeats, both peers have to opt in and use the value of 0.This is highly recommended against unless the environment is known to useTCP keepalives on every host.
Setting heartbeat timeout value too low can lead to falsepositives (peer being considered unavailable while it is notreally the case) due to transient network congestion,short-lived server flow control, and so on.
Several years worth of feedback from the users and clientlibrary maintainers suggest that values lower than 5 secondsare fairly likely to cause false positives, and values of 1second or lower are very likely to do so. Values within the 5to 20 seconds range are optimal for most environments.
Heartbeat frames are sent about every heartbeat timeout / 2seconds. This value is sometimes referred to as the heartbeat interval.After two missed heartbeats, the peer is consideredto be unreachable. Different clients manifest this differentlybut the TCP connection will be closed. When a client detectsthat RabbitMQ node is unreachable due to a heartbeat, it needsto re-connect.
It is important to not confuse the timeout value with the interval one.RabbitMQ configuration exposes the timeout value,so do the officially supported client libraries. However some clients might exposethe interval, potentially causing confusion.
Any traffic (e.g. protocol operations, published messages, acknowledgements) counts for a validheartbeat. Clients may choose to send heartbeat framesregardless of whether there was any other traffic on theconnection but some only do it when necessary.
Unless TCP keepalives are used instead with an adequately low inactivity detection period,deactivating heartbeats is highly discouraged. If heartbeats are deactivated, it will make timely peer unavailabilitydetection much less likely, which would pose a significant risk to data safety, in particular for publishers.
TCP contains a mechanism similar in purpose to the heartbeat(a.k.a. keepalive) one in messaging protocols and net ticktimeout covered above: TCP keepalives. Due to inadequatedefaults, TCP keepalives cannot be assumed to be suitablefor messaging protocols. However, with proper tuning they can beuseful as an additional defense mechanism in environments whereapplications cannot be expected to enable heartbeats or usereasonable values.
3a8082e126