Standard of Care: Reasonable But Not Perfect

The concept of standard of care in medicine has evolved over decades to include a legal definition of clinical care that is consistent with what a reasonable practitioner in the community would provide. In treating women of childbearing age who suffer from moderate to severe psychopathology, clinicians face challenging decisions, as they must consider the health and welfare of both the mother and the unborn child if the patient becomes pregnant. This case illustrates a difficult clinical situation in which, in our judgment, the treating clinicians provided treatment that met the standard of care. And yet the jury in the civil malpractice trial felt differently and awarded a substantial amount of money to the plaintiffs. How can we understand this verdict?

First, the distinction between optimal or “perfect” care and standard of care must be considered. Implicit in its definition is that the standard of care does not require either perfect care or the care that would be provided by an expert in the field. For the standard of care to be met simply requires that the threshold of reasonable care is provided. In this case, Dr. P provided a reasonable standard of care. The patient had severe schizoaffective disorder with multiple hospitalizations, was markedly dysfunctional and was a risk to herself and her family when ill, and had not responded optimally to many psychotropic medications in the past. Although other options, such as lithium, could have been considered, valproate was a reasonable alternative. (The atypical/schizoaffective features would predict a less than optimal response to lithium in this situation [1].) The patient was using an appropriate form of birth control—hormonal contraceptive patches (placing her risk of a pregnancy with a baby with a neural tube defect at less than 0.05%)—and she made it clear that she did not want to become pregnant during the next several years. Dr. P clearly informed the patient, on multiple occasions, of the teratogenic risk of valproate and documented these discussions in the chart. These warnings were repeated by Dr. M, the patient’s psychologist. The patient was capable of giving informed consent and understood the discussion. When the patient’s period was late the first time, Dr. P appropriately requested a pregnancy test, which was negative. Later, when Ms. A missed her period again, Dr. P again appropriately requested a pregnancy test, which the patient did not get in a timely manner. Despite urging by the treatment team for the test, it was 9 days until Ms. A finally got the result of the pregnancy test, which was positive, whereupon Dr. P appropriately discontinued the valproate immediately. Later, when an ultrasound examination revealed the presence of spina bifida in the fetus, Ms. A had multiple discussions with various members of her care team regarding her situation, including the option of terminating the pregnancy. Partly because of pressure from her husband, she declined a therapeutic abortion. She later gave birth to an infant with spina bifida and multiple other abnormalities.

This summary of events reflects thoughtful, involved, appropriate care of a woman with a severe psychiatric disorder, which seems to have met the standard of care despite the negative outcome of a disabled child. Yet the care was not perfect. While some clinicians may have made other treatment decisions, such as instructing discontinuation of valproic acid upon learning of a missed menstrual cycle or recommending two different types of contraception before prescribing valproic acid, Dr. P’s care was “reasonable,” falling within community standards. At a number of junctures, as was highlighted at the trial by the plaintiff’s expert witnesses, the care and the documentation of the care could have been different and probably better. As a resident, Dr. P did not discuss the initial prescription of the valproate with his attending faculty member; at that time, there was no documentation (and possibly no discussion) of alternative treatments, such as lithium. When Ms. A first missed her menstrual period and thought she was pregnant despite being on a reliable transdermal patch, Dr. P and the attending could have had a discussion with the patient to assess her thought process. Although the pregnancy test was negative, the missed menstrual cycle presented an opportunity for further clarification of Ms. A’s risks and an understanding of any possible mixed feelings regarding a pregnancy. When Ms. A missed her period a second time, Dr. P appropriately told her to obtain a pregnancy test. Yet he did not discontinue the valproate until the test results were returned. (It may be presumed that Dr. P assumed that the missed period did not reflect a pregnancy, as had occurred 6 months previously.) He did not walk her to the laboratory for the pregnancy test. Although Dr. M, the patient’s psychologist, also urged the patient to obtain the pregnancy test, Dr. P did not call the patient to more urgently remind her to get the test. Thus, the fetus was exposed to 9 extra days of valproate until the pregnancy test results were returned. (Whether this delay in discontinuing the valproate would have made a difference in the outcome is unclear. At trial, the two sides’ experts disagreed on whether the spinal cord defect would already have occurred at the time that Ms. A first noted her missed period.)

Complicating this case are the unclear boundaries inherent in a teaching institution. In this case, the responsibility of a psychiatric resident and the attending physician was, as is commonly the case, shared and, as a result, vague. Should it have been required that the attending faculty member be consulted before the valproate was prescribed? The attending knew about Dr. P’s decision later and seemingly did not object, but no note documented that discussion. Should the attending faculty member have been required to see the patient directly and then to document that visit and an opinion in the chart? Since the attending physician was aware of the valproate prescription later and did not change the treatment plan, it seems unlikely that the outcome would have differed if the attending psychiatrist had been more involved earlier. Thus, the outcome would not differ but the documentation would have been strengthened; however, the lack of documentation by the attending did affect the outcome of the trial. In a trial, “if it’s not documented, it never happened.” Although the resident did receive weekly supervision from the attending, it was not documented, leading to a false but difficult-to-refute impression of attending noninvolvement. The attending was aware of the resident’s handling of the second missed menstrual period, and if he had documented their discussion, thought process, and treatment plan in the medical record, it would have been beneficial at trial. This case suggests that it may be reasonable for training programs to set guidelines for attendings with regard to documentation of resident supervision and joint decision making. Furthermore, to minimize the risks of an adverse outcome, it seems prudent for major clinical decisions to occur in conjunction with the attending physician rather than in consultation with the attending after the fact.

It must also be acknowledged that different training programs have differing methods of dividing responsibilities between residents and attendings. Thus, no one national standard can be applied. Even within a single academic medical center, different clinics may have different standards, such that attending faculty in some clinics may see every patient at every visit for the entire visit (direct supervision), or for a portion of the visit, whereas in other settings (e.g., psychotherapy clinics), attending faculty supervise residents but do not see the patients directly. Dr. P’s supervision was asynchronous, occurring at a time that was detached from actual patient care. Currently, no national consensus on or standard for the structure of resident supervision in training institutions exists; perhaps uniformity for trainee supervision should be considered by the Residency Review Committee of the Accreditation Council for Graduate Medical Education. Factors to consider for standardization include how frequently attending supervision should occur (weekly? twice weekly?) and what type of documentation is reasonable. If the attending in this case had co-signed the resident’s note with the statement, “I have seen the patient with the resident and agree with the treatment plan that we discussed and jointly formulated,” would the outcome of the trial been affected?

Another conceptual issue raised at trial was the responsibility of the patient in contrast to that of the psychiatric treatment team. As one example, even though the patient was using a hormonal contraceptive patch, should the psychiatric treatment team have trusted that she was capable of using this method reliably? In treating a patient with a psychotic disorder, should we as psychiatrists be responsible for monitoring the patient’s adherence—or even capacity for adherence—to birth control methods that we do not prescribe? Despite the patient’s clear statements that she did not want to become pregnant in the near future, should this be considered before prescribing a teratogenic medication to a woman of childbearing age? In such cases it may be prudent at least to communicate concerns about birth control adherence with the patient’s obstetrician-gynecologist and encourage more frequent follow-ups.

The jury was clearly swayed by some combination of the imperfections in the patient’s care and, in all likelihood, sympathy for a psychiatrically ill woman caring for a severely developmentally disabled child. We cannot know how much of the verdict reflected sympathy and how much a critique of the psychiatric care she received, but both components assuredly contributed to the outcome. The sympathy factor is of course impossible to control and may overcome the strength of reasonable care that may have been provided. The lack of documentation by the attending psychiatrist in this case also had a significant impact on the outcome of the trial. Although the attending psychiatrist was involved in the decision making, the jury may have believed that the patient’s care was left solely up to the resident.

This case also highlights the significant risks associated with the use of valproic acid in pregnancy. These risks include spina bifida, developmental delay, atrial septal defect, cleft palate, hypospadias, polydactyly, craniosynostosis, and autism spectrum disorder (2, 3). At the time of this case, over 900,000 prescriptions of valproic acid were written annually for women between the ages of 15 and 44 (4). However, given the increased risk of significant adverse outcomes for the fetus, the use of valproic acid in women of childbearing age should be minimized. Furthermore, since nearly half of all pregnancies are unplanned (5), clinicians should routinely consider the possibility of an unplanned pregnancy when prescribing valproic acid to women of childbearing potential.

Complex cases sometimes go to trial. In this case, the plaintiff’s verdict was much less than what was asked. Although a jury may not be sophisticated enough to understand the complexity of psychiatric and medical issues, some cases may need to go to trial to defend against excessive settlement demands and verdicts. However, in cases that require an understanding of the challenges and subtleties of psychiatric care, a bench trial, in which the case is decided by a judge, without the assistance of a jury, may be warranted. When a lawsuit is filed, the plaintiff usually has the choice of a jury or a bench trial. Because juries are more likely to be influenced by a preponderance of evidence rather than burden of proof, are often swayed by a compelling story, and often grant larger awards, plaintiffs usually select a jury trial; therefore, most medical malpractice cases are decided by juries.

The overriding lesson in this case is that providing standard-of-care treatment may not be enough to prevent an adverse verdict in a malpractice case. As the cliché goes, bad cases make bad law. Therefore, we should be cautious about overinterpreting the results of this one case to change how we provide care, especially in an academic setting, in which issues of supervisory oversight are simultaneously amorphous and central. Yet, in high-risk cases—very ill psychiatric patients treated with medications in situations (e.g., childbearing age in a female patient) where there is substantial risk in both treating and not treating—special care should be taken to provide both optimal treatment and to document the thought processes underlying that treatment. For women of childbearing potential, the risk of an unplanned pregnancy needs to be considered. This case should, however, remind us of the uncertain outcome of difficult clinical cases, the risk of adverse outcomes, and the complexities of the legal system.