Practice Guidelines and Malpractice Litigation: A Two-Way Street

Practice guidelines are defined by the Institute of Medicine as systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances [1]. The Agency for Health Care Policy Research has committed substantial resources to guideline development, and numerous specialty societies and physician organizations, including the American Medical Association, have advocated the use of guidelines to improve quality of care. By some estimates, more than 1400 sets of guidelines are now available to the practicing physician [2]. Because they indicate a standard of care and are based on expert opinion, clinical practice guidelines bear on malpractice litigation [3]. A malpractice litigant must prove that he or she was injured by medical management that failed to reach the standard reasonably expected of the medical practitioner [4]. It follows that a physician who complies with a guideline that sets forth the standard of care could have a strong defense in a malpractice case; that is, he or she could use the guideline as exculpatory evidence. The guideline could range from being grounds for summary judgment for the defendant to playing a much less potent role as another part of the defense against a charge of negligence. Alternatively, failure to comply with a guideline might be evidence of negligence and, thus, would constitute inculpatory evidence. Again, the guideline might range from providing an inference of negligence to providing conclusive evidence of it. The interaction between guidelines and torts has been the subject of much analysis [5-7]. State and federal reformers who recommend experimentation with practice guidelines generally advocate the exculpatory use of guidelines [8]. For example, the State of Maine recently began an experiment [9] in which compliance with a guideline provides exculpatory evidence to help insulate a practitioner from a suit. Although it is limited to a few specialty areas and certified guidelines, this experiment is being closely watched [10]. Indeed, federal proposals contemplate much broader experimentation along these lines as part of a tort reform package [11, 12]. Reformers believe that guidelines can reduce litigation by encouraging compliance with the standard of care and that this will reduce health care costs by reducing the use of defensive medicine. Furthermore, the promise of lower rates of malpractice litigation will promote development of and greater compliance with guidelines, which will in turn improve the quality of medical practice and reduce costs associated with inappropriate care [13]. But those who believe that these benefits will result from reform typically assume that guidelines will be used only, or at least primarily, in one way: for exculpatory purposes. Little information is available, however, about how guidelines in general affect malpractice litigation today and, in particular, about the relative frequency of the use of guidelines for exculpatory as opposed to inculpatory purposes. Previous surveys have concentrated on published judicial opinions [14], which represent far less than 1% of all suits brought. The role that guidelines play in the overwhelming number of cases settled before trial is, at best, poorly understood. We undertook this two-part study to better document and address the use of guidelines in malpractice litigation. First, we reviewed litigation records at two professional liability insurance companies to investigate the use of practice guidelines by plaintiff and defendant attorneys. Second, we surveyed attorneys who litigate personal injury cases to assess their views on the role of guidelines. Our results suggest that practice guidelines will be used for both inculpatory and exculpatory purposes. Methods Review of Litigation Files We reviewed files of open and closed claims at two professional liability insurance companies in the United States, one in the northeast and the other in the intermountain west. In the states in which these companies are located, the statute of limitations for malpractice is 2 years and the mean period between date of injury and the date a claim was opened was less than 2 years. One company is physician-owned and provides primarily physician insurance throughout an entire state. The other is a large self-insurer of a group of hospitals and the physicians who practice at those hospitals. Because both open and closed claims were being evaluated, confidentiality concerns were substantial. Therefore, claims from both companies were pooled without any company identifiers attached. We requested 150 litigation files from each company. Because we thought that practice guidelines might play an important role in anesthesia and obstetrics claims, we requested all such claims that were opened as far back as 1987. We also requested a random sample of claims opened in the years 1989, 1990, and 1991. Our sampling strategy was dictated by our belief that practice guidelines were more likely to be used in more recent claims. On the other hand, we did not want the most recent claims because their files would often be incomplete. We reviewed the files using a survey instrument with primarily closed-ended, multiple-choice responses. The instrument was initially validated on a test set of claims files. The reviewers searched for references to guidelines in briefs, interrogatories, depositions, and other material in the files. They also noted the specific uses to which the guidelines were put and the parties invoking the guidelines, and they assessed the guidelines' weight in the litigation using structured, implicit judgment techniques. The reviews were done by three investigators. All of the reviews at one company and more than half of those at the other company were done by a physician who is also a lawyer and who has experience reviewing claims files. The rest of the reviews were split between an investigator who is both a physician and a lawyer and who is experienced in claim reviews and another who is a lawyer with substantial malpractice experience. All data points were keypunched, double-checked for errors, and entered into a SAS data set file. Survey of Attorneys To identify attorneys who devote a substantial amount of time to personal injury litigation, we searched a standard national directory of lawyers. All lawyers who had specified medical malpractice as an area of interest, a total of more than 9500, were identified. We then randomly selected 980 of these lawyers from all 50 states. In January 1993, we mailed each of these attorneys a questionnaire that was 4 pages in length. We requested information on the nature of the attorney's (or his or her firm's) practice; on the attorney's familiarity with the concept of practice guidelines as defined by the Institute of Medicine; and on the use of such guidelines in litigation. Most of the questions were closed-ended, but free-text responses were also captured. A second mailing was done 1 month after the first. All data points were keypunched, double-checked for errors, and entered into a SAS data set file. Analytic Methods Univariate proportions, such as the proportion of lawyers using practice guidelines, and their estimated standard errors were calculated using the SAS Proc Means program [15] (SAS Institute, Cary, North Carolina). The Fisher two-tailed exact test was used to examine associations between pairs of variables: for example, the association between the use of practice guidelines in a claim and the teaching status of the hospital in which the claim originated [16]. If one or both of the variables was an ordered categorical variable, the Mantel-Haenszel test statistic was used to test for association [17]. If both variables are ordinal, then using the Mantel-Haenszel test statistic is similar to testing whether a Spearman rank correlation coefficient equals 0 (under the null hypothesis of no association) [18]. These statistics and their corresponding P values were obtained using SAS Proc Freq. For a few tables, exact test statistics for associations between a pair of ordered categorical variables were used if the sample size of a contingency table was small; this was done using the STATXACT statistical package [19] (Cytel, Cambridge, Massachusetts). For binary outcomes, such as the use of practice guidelines, logistic regression was used for multivariate analyses: for example, to simultaneously identify significant predictors of the use of practice guidelines. For ordinal categorical outcomes, such as the average number of malpractice cases in which practice guidelines were used (categorized as 0, 1, 2 to 5, 6 to 10 and >10), proportional odds logistic regression was used for the multivariate analysis [20]. The logistic regression and proportional odds logistic regression estimates were obtained using SAS Proc Logistic software. The interpretation of the regression coefficients from the proportional odds logistic regression model is similar to the interpretation of the parameters from the usual logistic regression model. Results Review of Litigation Files Overall, 130 cases were obtained from one insurance company; the other 20 cases were not located in the time available for review. At the second insurance company, we reviewed 129 cases selected randomly from the 150 cases available. This produced a total of 259 claims. Many of the more recent claims were missing information about multiple entries. Seven files were too incomplete to be classified. The event year for claims ranged from 1976 through 1991; it was usually between 1986 and 1989 (n = 181; 69.9%) (Table 1). At the time of review, 42 (16.7%) of the claims had been closed with no payment and 111 (44%) had been settled with payment. Ten cases (4%) had gone to a jury and 89 (35.3%) were still open. In the three verdicts for plaintiffs, the awards were greater than $1 000 000. Two thirds of the cases settled were settled for less