If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to: Rashmi K. Sharma, MD, MHS, Division of Hospital Medicine, Northwestern University, 211 E. Ontario Street, 07-734, Chicago, IL 60611, USA.
Accurate documentation of inpatient code status discussions (CSDs) is important because of frequent patient care handoffs.
Objectives
To examine the quality of inpatient CSD documentation and compare documentation quality across physician services.
Methods
This was a retrospective study of patients hospitalized between January 1 and June 30, 2011 with a new or canceled do-not-resuscitate (DNR) order at least 24 hours after hospital admission. We developed a chart abstraction tool to assess the documentation of five quality elements: 1) who the DNR discussion was held with, 2) patient goals/values, 3) prognosis, 4) treatment options and resuscitation outcomes, and 5) health care power of attorney (HCPOA).
Results
We identified 379 patients, of whom 235 (62%) had a note documenting a CSD. After excluding patients lacking a note from their primary service, 227 remained for analysis. Sixty-three percent of notes contained documentation of who the discussion was held with. Patient goals/values were documented in 43%, discussion of prognosis in 14%, treatment options and resuscitation outcomes in 40%, and HCPOA in 29%. Hospitalists were more likely than residents to document who the discussion was held with (P < 0.001) and patient goals/values (P < 0.001), whereas internal medicine residents were more likely to document HCPOA (P = 0.04). The mean number of elements documented for hospitalists was 2.40, followed by internal medicine residents at 2.07, and non-internal medicine trainees at 1.30 (P < 0.001).
Conclusion
Documentation quality of inpatient CSDs was poor. Our findings highlight the need to improve the quality of resident and attending CSD documentation.
Cardiopulmonary resuscitation is the default treatment option for hospitalized patients with cardiac arrest. As a result, assessment and documentation of patient preferences regarding resuscitation is needed to ensure that patients receive care concordant with their preferences.
Experts suggest that when conducting a code status discussion (CSD), physicians discuss resuscitation in the context of the patient's clinical condition, values, and goals and provide information about treatment options and expected outcomes.
making the quality of documentation about important patient decisions, such as resuscitation, more important than ever. However, little is known about the quality of inpatient CSD documentation and whether documentation differs across physician services.
The objectives of this study were to evaluate the quality and content of inpatient CSD documentation based on five key quality elements and to compare the quality of CSD documentation between internal medicine residents, hospitalists, and non-internal medicine trainees.
Methods
Study Design, Participants, and Setting
We conducted a retrospective medical records review of patients admitted to our 894-bed urban tertiary care hospital, between January 1, 2011 and June 30, 2011. Patients are admitted under the care of a primary service (e.g., internal medicine, hospitalist, general surgery, obstetrics, and gynecology). Patients admitted to medicine and subspecialty services are cared for by teams of resident physicians under the supervision of an attending physician. Patients admitted to the hospitalist service receive care from a hospitalist faculty member without resident physician coverage. All physician documentation is entered into the electronic medical record.
Procedure
We used the Northwestern University Enterprise Data Warehouse (EDW)
to search for all patients with a do-not-resuscitate (DNR) order written or canceled at least 24 hours after admission. The EDW is a computer-based system that uses structured query language syntax to obtain demographic and clinical information derived from inpatient and outpatient electronic medical record systems. We restricted our search to orders written or canceled at least 24 hours after admission to capture patients with whom an explicit CSD had likely occurred rather than those who had existing documentation of code status preferences on admission (e.g., patients who had a DNR order on a previous admission). We then used an electronic text search query through the EDW to identify all clinical notes written within 24 hours before and 24 hours after the DNR order. The EDW is able to retrieve full note text and identify the name, title, and departmental affiliation of all note authors.
One study author (A. T.) reviewed all physician notes generated from the search. Notes written by consultant physicians and medical students were excluded from analysis. If a patient had multiple notes documenting a CSD, either from the same or different physicians, we included the first note written after the DNR order. If there were primary service notes before the DNR order that referenced code status but none afterward, we included the note that occurred closest to the time of the DNR order. For patients who had a DNR order and multiple hospitalizations during the study period, we only included data from the index hospitalization.
After the final list of included notes was identified, we categorized them into three groups based on note author: 1) internal medicine intern or resident, 2) internal medicine hospitalist attending physician, and 3) non-internal medicine physician. Non-internal medicine physicians included non-internal medicine residents (e.g., anesthesiology, surgery, neurology) and subspecialty fellows. The six-month study period was deemed sufficient to provide a sample size of 73 per group, which would have 80% power to detect significant differences in documentation quality ratings between physician specialties at an alpha of 0.05. The Northwestern University Institutional Review Board approved the study.
Main Measures
Patient Characteristics
We abstracted the following sociodemographic data from the EDW for each patient: age, sex, and race/ethnicity. Clinical data from the EDW included the date and time when the DNR order was written or canceled, hospital lengths of stay, the presence of a palliative care consultation, prior admission to the study hospital in the last six months, referral to hospice, and in-hospital death. Severity of illness was assessed by the all-patient refined diagnosis-related group (APR-DRG) complexity score, which ranges from 1 to 4 indicating minor to extreme complexity. We dichotomized the APR-DRG score as a 4 vs. less than 4 given the skewed distribution of this measure.
Quality of Documentation
We identified five key CSD quality elements based on published guidelines on CSD content
and input from three clinical experts in internal medicine, geriatrics, and palliative medicine, which was supplemented by items from a checklist that we previously developed and evaluated as part of a resident CSD skills training program for discussion of code status.
We developed a chart abstraction tool to assess documentation of five quality elements: 1) documentation of who the discussion was held with (“discussant”), 2) discussion of patient goals and/or values, 3) discussion of prognosis, 4) review of treatment options and/or resuscitation outcomes, and 5) identification of a health care power of attorney (HCPOA).
Abstraction rating and scoring parameters were determined in advance of the study. Study investigators and three clinical experts reviewed a sample of 20 CSD notes to reach agreement on precise definitions for each quality element. We defined physician documentation of the discussant as documentation in the note of who participated in the CSD (e.g., “discussed code status with the patient” or “discussed with the patient's son”). Documentation of patient goals and/or values was defined as documentation about the patient's objectives and what would be important to the patient (e.g., “patient wants to go home” or “patient would like to focus on being comfortable”). We considered documentation of prognosis present if there was specific identification of prognostic information (e.g., “poor patient prognosis” or “life expectancy of hours to days”). Documentation of resuscitation outcomes and/or treatment options included anticipated outcomes of resuscitation or identification of end-of-life treatments discussed with the patient or family (e.g., “patient does not want intubation” or “unlikely return of physical function after resuscitation”). Finally, we considered identification of an HCPOA if the name of the HCPOA or surrogate was explicitly stated in the note (e.g., “POA: Bill Smith”). We also included instances where the physician documented that the patient was asked if he or she had an HCPOA and received a negative response.
Once consensus on rating was obtained, one coder (A. T.) abstracted data on the presence or absence of the five quality elements in each abstracted note using the chart abstraction tool. A second coder (R.K.S.) reviewed a subset of 50 notes to assess interrater reliability using the kappa coefficient.
Statistical Analysis
We used standard descriptive statistics to evaluate characteristics of the study sample. For the subset of patients who had a CSD note, we calculated frequencies for the presence of each of the five quality elements and used Pearson Chi-squared tests to evaluate the association between the presence of each quality element and physician service. To evaluate the association between patient demographics (e.g., age, sex, and race) and clinical characteristics (e.g., severity of illness, lengths of stay, palliative care consultation) and the presence of each quality element, we used analysis of variance for continuous variables and Chi-squared tests for categorical variables. All data analyses were conducted using Stata, version 11.0 (StataCorp, College Station, TX).
Results
Sample Characteristics
Three hundred seventy-nine patients had a DNR status change reflected in the presence of a new or canceled DNR order 24 hours after hospital admission. Of these patients, 235 (62%) had a note documenting a CSD during the 24 hours before and/or 24 hours after placement of the order. We were left with 227 patients after excluding eight patients who had a note from a consulting service but did not have a note from the primary service documenting a CSD. As shown in Table 1, characteristics of patients with and without CSD notes were similar. Approximately half of the patients were females and older than 65 years. Most had been admitted within the past six months, and almost 70% had an APR-DRG score of 4 indicating high severity of illness. There were significant differences in physician service for patients who had documentation of a CSD compared with those who did not (P < 0.05) with most patients who did not have documentation being on an internal medicine resident service.
Table 1Characteristics of 379 Patients Admitted Between January 1 and June 30, 2011 With a New or Canceled DNR Order by Documentation of Code Status Discussion
Patient Characteristics
Total Sample (N = 379 Patients)
No CSD Documentation (N = 152 Patients)
CSD Documentation (N = 227 Patients)
Age 65 yrs and older (%)
56.5
55.3
57.3
Sex (% female)
51.7
46.7
51.4
Race (%)
White
61.7
63.2
60.8
Black
24.3
23.7
24.7
Other
14.0
13.2
14.5
Lengths of stay (days), median (interquartile range)
Sixty-nine notes (30%) were written by 34 internal medicine residents. Seventy-two notes (32%) were written by 31 hospitalists. Eighty-six notes (38%) were written by 61 non-internal medicine residents. A graphic display of the number of quality elements present for CSD notes is shown in Fig. 1. Twenty-two percent of notes (N = 50) did not contain a single quality element and received a score of 0. On the five quality measures by which the notes were assessed, almost 50% of notes scored either a 2 (24%, N = 55) or a 3 (24%, N = 54). Only 2% of notes (N = 4) contained all five quality elements.
Fig. 1Number of elements present (range 0–5) for hospital notes documenting a code status discussion (N = 227).
Table 2 displays CSD quality elements by physician service. Interrater reliability was high for each quality element. Overall, physicians documented who they discussed code status with (discussant) in 63% of notes, patient goals/values in 43%, prognosis in 14%, treatment options or resuscitation outcomes in 40%, and HCPOA in 29%. Physicians documented both a full code and DNR status for the same patient in eight notes.
Table 2Quality Elements Documented in Hospital Code Status Discussion Notes by Physician Service (N = 227)
Notes written by hospitalists were more likely to include documentation of the discussant (86%) than those written by internal medicine residents (71%) and non-internal medicine trainees (36%; P < 0.001). Hospitalists were also more likely to document patient goals/values (63% of notes) than internal medicine residents (42%) and nonmedicine trainees (28%; P < 0.001). Internal medicine residents were more likely to document HCPOA (41%) than hospitalists (26%) and nonmedicine trainees (22%; P = 0.04). The mean number of elements documented for hospitalists was 2.40, followed by internal medicine residents at 2.07 and non-internal medicine trainees at 1.30 (P < 0.001).
Factors Associated With Documentation Quality
Using Chi-squared analyses, we evaluated the association between patient factors (e.g., demographics and clinical characteristics) and documentation of each of the quality elements. We found that older patients (older than 65 years) were more likely to have documentation of HCPOA than younger patients (35% vs. 21%; P = 0.02) but less likely to have documentation of prognosis (7% vs. 24%; P < 0.001). We also found that black patients were more likely to have documentation of prognosis (19% vs. 9%; P = 0.02) than white patients. Patients who had a palliative care consultation were more likely than patients who did not have a consultation to have their primary service document the patient's goals/values (57% vs. 29%; P = 0.001), treatment options or resuscitation outcomes (53% vs. 27%; P = 0.001), and prognosis (19% vs. 9%; P = 0.02) but less likely to document HCPOA (21% vs. 37%; P = 0.008). There was no significant association between patient sex, severity of illness, lengths of stay, or prior admission within six months and documentation of any of the quality elements.
Discussion
This study demonstrates that the quality of inpatient CSD documentation is poor. We found that a quarter of notes did not include any of the five key quality elements and another quarter only had a single element. Furthermore, only a minority of notes included documentation about prognosis (14%) and HCPOA (29%). These findings confirm results from a study of terminally ill cancer patients in which documentation of discussions about prognosis was limited and infrequent.
Such poor documentation is especially troubling because physician-patient discussion about end-of-life preferences has been associated with higher quality of life
Although overall CSD documentation was poor, we did identify unique differences in quality between specific physician services. Specifically, hospitalists were more likely to document discussant and patient goals/values than internal medicine residents and non-internal medicine trainees. There may be several possible explanations for the difference in CSD documentation quality between physician services. The additional clinical and CSD experience of hospitalists may positively affect documentation by nature of the volume of exposure to these situations. In addition, frequent handoffs between hospitalists
may promote clear documentation of elements such as discussant and goals/values to avoid confusion in care planning. The solitary nature of being a hospitalist, when compared with the team dynamic of resident services, may also contribute to a perceived pressure for improved documentation because of the lack of multiple team members documenting on the same patient (i.e., an intern may feel like his or her senior resident or attending will document elements of the CSD and therefore not include these elements in his or her note).
We also found that internal medicine residents were more likely to document HCPOA (41%) than nonmedicine trainees (22%; P = 0.04) and document a greater number of elements overall (2.07 vs. 1.30; P < 0.001). Variations in the culture of medicine may explain some of the differences we found between medicine and nonmedicine services, with perhaps greater emphasis being placed on documentation of HCPOA in medicine training programs compared with nonmedicine programs. Additional factors that could help explain these differences in documentation quality include differences in the time available for documentation and patient volume between services. The lower number of key quality elements on non-internal medicine services may also reflect differences in perceived responsibility, with the documentation of CSD elements perceived as the responsibility of other services (i.e., internal medicine, palliative care, social work, pastoral care) and not the primary service. The increased documentation of discussant, goals/values, and HCPOA compared with other elements may also reflect a focus on discharge care planning on internal medicine services, whereas prognosis and resuscitation outcomes/treatment options may not be perceived as crucial to the care plan.
We also found that some patient characteristics, such as younger age and black race, were associated with documentation of prognosis in particular. Furthermore, the presence of a palliative care consultation was associated with increased documentation of several of the quality elements. Because we excluded consultant notes from our analysis, and thus all notes written by the palliative care service, our results suggest that the presence of a palliative care consultation may improve documentation of other physicians caring for the patient particularly with regard to documentation of the patient's goals/values, resuscitation outcomes, and prognosis. Whether documentation quality improved because physicians had a higher standard to emulate, or added elements from the consultant note to their own, is unclear. The presence of a palliative care consultant in a given CSD may also improve the overall quality of the discussion, thereby resulting in improved documentation by the primary service. Of note, the high rates of palliative care consultation noted in our study sample likely reflect the high severity of illness of patients having CSDs. Physicians may be more likely to discuss code status with patients who they perceive at high risk of arresting in the hospital.
Our study has several limitations. Data were restricted to a single academic medical center and a relatively small number of clinical events, and we were not able to evaluate the association between quality of documentation and other clinical outcomes. We also chose to exclude patients who had a DNR order within the first 24 hours of admission so that we could focus on patients who were more likely to have had an explicit CSD with their physician. We hypothesized that many DNR orders placed on the day of admission would be based on CSDs from previous admissions and re-entered into the electronic record by the admitting physician. We suspect that these notes would thus be of poorer quality and would bias our results toward even lower rates of documented quality elements. Additionally, we did not measure resident or hospitalist experience conducting CSDs or exposure to training in palliative care, the results of which could potentially affect frequency and quality of CSD documentation. It is also important to note that our study reflects the quality of CSD documentation, which may or may not reflect the quality of the actual CSD itself. Finally, we were not able to cross-check physician service for patients who did not have documentation of the CSD and relied on the admitting service listed in the EDW.
In conclusion, our findings highlight the need to improve the quality of inpatient CSD documentation. With increasing numbers of handoffs between medical services and the resulting need to effectively communicate patient preferences for end-of-life care between physicians, it is imperative to develop interventions to ensure high-quality CSD documentation. Potential options include CSD documentation training as part of communication skills education and the use of documentation templates. Further study is needed to assess the impact of such interventions on CSD documentation and downstream patient care.
Disclosures and Acknowledgments
Dr. Sharma is supported in part by grant number K12 HD055884 from the Eunice Kennedy Shriver National Institute of Child Health & Human Development. The authors declare no conflicts of interest.
References
Wenger N.S.
Phillips R.S.
Teno J.M.
et al.
Physician understanding of patient resuscitation preferences: insights and clinical implications.
00136-5&id=gr1.jpg){kind=link}