|Home | About | Journals | Submit | Contact Us | Français|
Wendell F. Rosse, former chief of Hematology and Medical Oncology at Duke University (Durham, NC), taught all of his students that it is the physician and not the patient who serves as historian in the construction of a medical record, because it is the physician, with his or her team, who collects, collates, interprets, and records the details of the patient's history.
There is a lot of information to be placed into the schematized traditional format for recording and presenting this history, including the patient's complaint, information relevant to that complaint, and the review of other autobiographic health information provided by the patient both spontaneously and in response to questioning. In addition, findings from a physical examination are provided, as are results of a listing of test data that are becoming increasingly complex and often difficult to find. It is no surprise that this process sometimes ends with missing or erroneous data.
Clearly, these data may or may not be relevant to clinical decision making. For example, in this issue of Journal of Oncology Practice, although Abernethy et al1 note that sex was missing or unknown in 17% of the patient medical records they examined, it is unclear whether sex has any impact on treatment decisions in colorectal cancer. Information regarding health insurance and race is deemed relevant by Abernethy et al as well, and although there are critical issues linking socioeconomic status and race with access to care,2 these two patient characteristics also seem particularly irrelevant in deciding appropriate care for patients with colorectal cancer. In addition, it is difficult to understand the finding that practice type was missing or unknown in 131 (26%) of 499 patient medical records suitable for analysis. One can only imagine that either the data were not transmitted or the definitions of practice type were unclear. The treating physicians must have been known and recorded, if only as signatures or electronic log-ons.
Evaluation of quality of care on the basis of written records is fraught with hazard. Donabedian,3 a pioneer in health services research, codified the problems in an article in 1966, which was reprinted in 2005. He described office records as “disappointingly sketchy,” and he noted the lack of an adequate record did not necessarily demonstrate that inadequate care had been administered. In addition, he noted that secondary reporting of diagnostic information could itself be erroneous, either because the reporting was inconsistent with the original or because the original interpretation itself was wrong.
This is critically important when examining staging information in oncology, when accurate stage determination can only be assigned after a complete review of all relevant data. This is not to say that primary review of pathologic information, a hallmark of oncology care,4 can be ignored. Abernethy et al1 report that pathologic information was included in the medical records of 66% of the entire population of 499 patients and in 84% of the final 61 eligible patients; however, these rates are notably lower than those observed over time in the ASCO Quality Oncology Practice Initiative data, in which pathologic information was consistently present in more than 95% of medical records.5
Others have noted more pernicious problems with written records. For example, Luck et al6 compared quality metrics developed on the basis of medical records with those developed on the basis of patient reports by trained-actor patients for four chronic medical conditions, including lower back pain, chronic obstructive pulmonary disease, diabetes, and coronary disease. They defined a false positive as presence in a medical record of something that did not actually happen (as reported by patients immediately after their physician encounters). In 272 (19%) of 1,456 omissions of care reported by patients, the medical records reported that correct care had been administered. More encouraging for patients (if alarming to physicians, who are likely to be judged and rewarded on the basis of their patients' medical records) were omissions of records of proper care in 918 of 3,084 events in which correct care had in fact been delivered.
Electronic medical records (EMRs) have been heralded as an answer, because EMRs can be programmed to stop the caregiver until and unless particular data elements are recorded. Abernethy et al1 did find higher rates of information available in those situations in which EMRs were used. But EMRs are not a panacea. Erroneous information takes on a life of its own through the ease of cutting and pasting, allowing the propagation of “medical record lore.” The use of default templates allow facile collection of often overwhelmingly numerous and sometimes false records, which obscure more important and pertinent observations.7
How can we fix this if we as physicians are to be subject to review by others? First, we need to move toward measurement of outcomes as a more important standard.8 Second, we should be sure that relevant patient comorbidities are available to those who are passing judgment, as has been done in cardiothoracic surgery9 and is being done in the ASCO Breast Cancer Registry Pilot. Until this can be done, process-based measures of quality need to be developed with broad representation of and input from physician, patient, and payer stakeholders, as is planned for the ASCO Quality Oncology Practice Initiative expansions this year. Most importantly, we need to insist that payers recognize the difficulty, effort, time, and skill required to piece together the histories of our patients.
The author indicated no potential conflicts of interest.