Adverse Events

INTRODUCTION

The idea that problem medical care can be identified through poor patient outcomes that are unexpected is behind the "occurrence screening" and "incident reporting" systems that have been implemented in almost all U.S. hospitals. Touted as early warning systems for hospital administrators, occurrence screening and incident reporting systems grew out of the malpractice crisis of the mid1970s, when institutions desperately began to seek ways to limit their liability. Exactly what constitutes an occurrence or an incident varies widely among institutions. Although most reporting systems use patient outcomes as criteria to screen for occurrences or to define incidents, some also use criteria related to the process of care. The single thing that all the reporting systems have in common is that they are used by hospitals only as a first step for finding poor-quality medical care. In many cases, the occurrence of adverse events may result from factors other than poor quality. Thus, to establish a link between the quality of hospital care and adverse events, hospital cases identified by the reporting systems must be followed up with more thorough investigation and interpretation by medical advisers.

In the early 1970s, Rutstein and his colleagues proposed counting "sentinel health events," or cases of unnecessary diseases, disabilities, and untimely deaths, to monitor the quality of medical care (546). Working with numerous specialists, these researchers developed a list of specific conditions for which adverse outcomes-whether caused by commission or omission-should never occur, such as death from tuberculosis.

Specific criteria for reporting adverse incidents across all conditions were first developed in 1976 in the California Medical Insurance Feasibility Study (432). That study, sponsored by the California Medical Association and the California Hospital Association, used general outcome criteria to screen more than 20,000 patient charts from 23 hospitals for adverse events that might result in litigation for malpractice compensation.

The 20 "potentially compensable events" developed by physicians and medical audit experts in the 1976 California study later became the basis for "occurrence screens," adapted and modified for use by individual institutions. An adaptation of the general outcome criteria that was developed by Medical Management Analysis is shown in table 5-1 (154). The outcome criteria in the table, now used in more than 200 U.S. hospitals, cover all aspects of hospitalization and are generally used to screen every patient record during the patients' hospital stay (290).

Among the common adverse events used as criteria in most occurrence screens are deaths, nosocomial (hospital-acquired) infections, unusually long lengths of stay, and unscheduled procedures, readmissions, or transfers. The use of deaths as a criterion may be limited to cases where death is a statistically rare outcome for the procedure, condition, or diagnosis-related group or is in some other way unexpected. In most hospitals, cases with adverse events identified by an occurrence screen are subsequently reviewed in depth for possible problems related to the quality of care. Almost all hospitals adapt occurrence screens for their own particular needs, for example, adding suitable clinical indicators developed at the departmental or service level. The use of occurrence screens in a hospital is usually part of the hospital's quality assurance program and therefore directly linked with existing peer review endeavors.

It is not known how many U.S. hospitals currently use occurrence criteria to screen their patient populations for adverse events. Increasingly, insurance companies are requiring hospitals to use occurrence screens as a condition for underwriting the medical malpractice insurance of affiliated physicians (420). The Joint Commission on the Accreditation of Healthcare Organizations (JCAHO) encourages the use of specific criteria to select cases for review in hospitals' quality assurance programs, yet it gives ample leeway in

Criterion 4: Unplanned removal, injury, or repair of organ structure during surgery or other invasive procedure, or vaginal delivery. Criterion 5: Unplanned return to operating room, delivery room, or other special procedures room on this admission.

Criterion 6: Surgical and other invasive procedures which do not meet criteria for necessity and appropriateness.

a. Diagnostic tissue-pathology report does not match preoperative diagnosis.

b. Nondiagnostic or normal tissue removed and medical staff criteria for necessity or appropriateness not met.

c. No tissue removed and medical staff criteria for necessity and appropriateness not met.

d. Other.

Criterion 7: Blood loss excessive or blood/blood component utilization which is unjustified, excessive, results in patient injury, or is otherwise at variance with professional staff criteria.

a. Excessive blood loss occasioned by iatrogenic bleeding or anemia with or without transfusion.

b. Transfusion of blood or blood components not clinically indicated. c. Transfusion reaction.

d. Other.

Criterion 8: Nosocomial infection (hospital-acquired infection).

Criterion 9: Drug/antibiotic utilization which is unjustified, excessive, inaccurate, results in patient injury, or is otherwise at variance with professional staff criterion.

a. Does not meet professional staff criterion for appropriateness. b. Inadequate/excessive/inappropriate/inaccurate dosage or timing. c. Drug or contrast material reaction/interaction. d. Other.

Criterion 10: Cardiac or respiratory arrest/low Apgar score. Criterion 11: Transfer from general care to special unit.

Criterion 12: Other patient complications.

Criterion 13: Hospital-incurred patient incident.

a. Falls, slips, patient accident.

b. Intravenous problems, such as calculation errors, overloads, or infiltrations.

c. Skin problems, such as rash, threatened or new decubitus ulcer.

d. Equipment failures/malfunctions.

e. Other incidents, such as procedural errors, electrical shock or burn, actual or attempted suicide, and lost or damaged property.

Exceptions: Specific instructions may be developed by the clinical departments concerning expected admissions for chronic conditions managed in the outpatient setting.

Exceptions:

• Complication or incomplete management occurred at another hospital not associated with this hospital or involved a practitioner who is not on this medical staff.

• Planned admissions for secondary procedures needed to complete treatment.

Exceptions:

• Emergency procedures where the patient was unable and the family or legal guardian unavailable to sign the consent.

• Life-threatening problems found and addressed during surgery.

Exceptions: None.

Exceptions: Planned second procedure or second stage of a procedure planned prior to first procedure.

Exceptions: As developed by the medical staff.

Exceptions: As developed by the professional staff.

Exceptions: Infection acquired outside this hospital, clinic, or home health care setting and did not involve any member of this medical staff.

Exceptions: As developed by the professional staff.

Exceptions: None.

Exceptions: Transfer scheduled prior to surgery or other special procedure.

Exceptions: None.

Exceptions: None.

the degree of specificity. There are certainly wide disparities in the occurrence screens used by hospitals. All or samples of patient populations can be screened for occurrences either during the patients' hospital stay or retrospectively after discharge. Hospitalwide, "generic" screens can be applied equally across all patients, or detailed service-specific criteria devised for similar sets of patients. Screens can be computerized too, but the level of patient information detailed in the screens usually requires the review of patients' medical records by specially trained personnel in all but a few highly computerized hospitals.

Incident reporting systems, though often overlapping with occurrence screens and also growing out of concerns about rising malpractice liability, tend to be organized and operated somewhat differently from occurrence screens. Incident reporting systems are organized directly by the hospital administration (rather than being part of a hospital's quality assurance program) and tend to be operated independently of the medical record or other existing information systems. Typically, as part of risk-management programs, hospital personnel (most frequently nurses) complete forms when they observe an adverse event, and the forms are reviewed centrally by a hospital administrator/risk manager. The definition of an "incident" is often left to the discretion of the frontline health professionals who deal with patients. Most commonly, adverse events such as

patient falls, medication errors, equipment failures, and commission of procedure or treatment errors are considered incidents. Reliance is placed on educating nurses, physicians, and other health care workers to recognize problems and report them.

Because health care personnel use their judgment in reporting incidents, it is more likely that incidents reflect quality-of-care problems than do the adverse events that are initially identified by occurrence screens; screening systems are expected to identify substantial numbers of false positives. Although reported incidents might therefore be viewed as being one step closer to identifying poor-quality care than are occurrences picked up in screens, further investigation of incidents is also necessary. First, an incident may not have been caused by negligent medical care; for example, a patient fall may have resulted from the patient's own carelessness. Second, an incident may not have had an important impact on the patient; for example, even though a medication has been administered incorrectly, a patient may suffer no ill effects.

Almost all hospitals have incident reporting systems, but the quality and reliability of reporting in these systems vary enormously across institutions. Currently, eight States1 and the Veterans

'The eight States are Alaska, Florida, Kansas, Maryland, Massachusetts, New York, Rhode Island, and Washington (290).