- Email: [email protected]
Preformatted charts improve documentation in the emergency department
ORIGINAL CONTRIBUTION
documentation; medical records
Preformatted ChartsImproveDocumentationin the EmergencyDepartment From the University of Pennsylvania School of Medicine;* the Emergency
Tanya Humphreys,MD* Frances S Shofer, PhD*" Sheldon Jacobson, MD*§
Servicest and Obstetrics and
Christos Coutifaris, MD, PhD~
Gynecologyt Departments,
Annette Stemhagen, DrPHt§
Study objectives: Todetermine if the use of programmed charts with complaint-specific entry criteria results in improved documentation of patient encounters and better clinical outcome. Design: Prospective study.
Hospital of the University of
Setting: Emergency department of an urban university hospital.
Pennsylvania; and the Section
Type of participants: Female patients presenting to the emergency
of General Internal Medicine,
department with gynecologic complaints of abdominal pain, bleeding, or vaginal discharge.
Department of Medicine, University of Pennsylvania, Philadelphia.§
Interventions:
Received for publication November 26, 1990. Revision
Measurements:
received September 23, 1991. Accepted for publication January 28, 1992.
Presented at the Societyfor Academic Emergency Medicine Annual Meeting in Minneapolis,
Minnesota, May 1990.
Programmed and blank charts were provided randemly for physicians in the ED.
Chart scores based on documentation criteria for patient history, physical examination, laboratory studies, diagnosis, and discharge instructions and patient outcome scores of 0% to 100% based on the persistence of their complaints at the time of the followup interview. Main results: Overall documentation of history, physical examination, and laboratory studies was more complete on programmed charts than on blank charts (81.1% vs 71%, P< .0001). The patient history portion of the charts was found to benefit the most from the use of programmed charts (74.8% vs 60.1%, P< .0001). Although programmed charts demonstrated better documentation, there was no statistically significant correlation with patient outcome parameters or with patient satisfaction with the quality of medical care. However, more patients whose physicians used programmed charts were satisfied with their physicians' explanations of their problem (X~ = 5.2, P< ,02). Conclusion: Programmed charts improve documentation by facilitation of the documentation process and allow more time for patientphysician interaction. Quality of documentation alone, however, is not a reliable indicator of patient outcome or of the quality of care received. [Humphreys T, Shofer FS, Jacobson S, Coutifaris C, Stemhagen A: Preformatted Charts improve documentation in the emergency department. Ann EmergMad May 1992;21:534-540.]
MAY 1992 21:5 ANNALS OF EMER6ENCY MEDiCiNE
53 4 / 7 1
PREFORMATTED
CHARTS
Humphreys et al
INTRODUCTION
Preformatted problem-oriented patient encounter charts (programmed charts) have been used sporadically by emergency physicians as a means of simultaneously increasing the accuracy of documentation and reducing the amount of time spent documenting patient care. These charts, which provide a complete list of important documentation criteria for different medical complaints, also may be a valuable teaching tool for housestaff training. The relationship between the quality of the medical records and patient outcome remains controversial. There is no conclusive evidence that the completeness of information recording in medical records reflects the quality of medical care given. Many studies, in fact, suggest that medical records alone are not a reliable indicator of patient care or health outcome, x-a However, theissue of accurate documentation is an important one, as medical records are used frequently by third-party payers to validate claims and by hospital administrators and lawyers to evaluate physician performance for quality assurance audits and malpractice
Figure 1. Programmed chart
]
~ ' ~
•
U
•
EMERGENCYSERVICES MEDICAL RIPORT
'
F
I"
s.u,s.:
~G
(_ E ~ R C e N C V S~RV,CeS O E P A " ~ E N T
72/535
F
/ ........
GYN/ 08 UTI
I ....
P,~.~:
o ua,s,m,
o ~
suits. More important, medical records are a vital means of communication between patient care providers. Completeness of documentation represents a special challenge for emergency physicians who work in high-volume, high-acuity settings with competing demands of patient care and administrative responsibilities. We sought to establish the ability of programmed charts to improve documentation in the El) setting and to evaluate the effect, if any, of preformatted charts on the clinical outcomes of patients.
MATERIALS
AND
METHODS
Programmed charts were designed by the attending ED staff at the Hospital of the University of Pennsylvania for recording obstetric and gynecologic evaluations in the El) (Figure 1). These charts are preformatted with specific prompts for information pertaining to patient history and physical examination. The study was conducted in the ED of the Hospital of the University of Pennsylvania, a large urban teaching hospital with 45,000 ED visits annually. Approximately 15% of these visits are for obstetric and gynecologic complaints; of these, 15% to 20% are admitted. Between February 1989 and August 1989, either standard blank charts or preformatted programmed charts were provided on alternate weeks by ED secretaries to internal medicine housestaff and emergency attending physicians for the documentation of patient encounters. Approximately 5% of completed programmed and blank charts of patients between the ages of 15 and 45 years seeking care in the ED for obstetric or gynecologic problems were selected randomly for evaluation of completeness. Relevant charts were identified and retrieved through the use of the El) daily log by a trained volunteer. Because of the medical records practices in our hospital, charts of patients who were admitted were not available for inclusion in this study. Charts that were illegible also were not included for evaluation. All patients were blinded as to chart type used. A panel of three obstetrician/gynecologists and one emergency physician developed a set of documentation criteria for patient history, physical examination, laboratory studies, diagnosis, and discharge instructions. Separate documentation criteria were established by group consensus for five categories of patient complaints frequently encountered in the El): pregnant and nonpregnant women with abdominal pain, pregnant and nonpregnant women with vaginal bleeding, and nonpregnant women with vaginal discharge. Then, each individual documentation criterion was assigned a weighted value (1 through 3) based on its relative importance with respect to the specific complaint category as determined by the panel. Examples of documentation criteria for nonpregnant women with abdominal pain included specific historical information, physical examination findings, and laboratory criteria (Figure 2). Similar •
ANNALS 0FEMER6ENCYMEDICtNE
21:5
MAY1992
PREFORMATTED CHARTS f l u m p h r e y s et al
criteria were applied to charts of the other categories of chief complaints. Each completed c h a r t (programmed and blank) was reviewed by a single reviewer (not p a r t of the original panel) using a s t a n d a r d evaluation sheet formatted for each complaint category, and a c h a r t score was calculated based on the presence or absence of criteria deemed necessary for complete documentation. Then, the percentage of the total possible score was calculated for each c h a r t to allow greater ease of comparison among categories. Attempts were made to contact by telephone all patients whose charts had been collected for evaluation. A trained Figure2. Documentation criteria for nonpregnant women with abdominal pain
Criterion Weight 3 3 2
2 3 3 1 1 1 3 2 3 3
2 3 3 3 3
History Location of pain Duration of pain Radiation Where did pain start? Has it moved? Bowel symptoms or gastrointestinal complaints (nausea, vomiting, diarrhea) Last menstrual period History of fever Other medical problems, including surgical history Medications Birth control pill use Allergies Vaginal discharge Sexual activity Reproductive history (gravity, parity, history of abortions, past pregnancy problems, pelvic inflammatory disease history) Physical Examination Cardiopulmonary Pelvic Rectal Flanks (costovertebra[ angle tenderness) Abdomen Laboratory Pregnancy test -- negative Urinalysis (microanalysis or sedimentation and dipstick) CBC Cultures If there is a mucopurulent discharge Cervical Vaginal wet mount (saline) Consultation and Diagnosis If consult needed and is called (no extra credit if mere than one consult is called)
MAY1992
21:5 ANNALS 0F EMERGENCY MEDICINE
Yes [ ] [ ]
No [ ] [ ]
[ ] [ ]
[ ] [ ]
[ ] [ ] [ ]
[ ] [ ] [ ]
[ [ [ [ [ [
[ [ [ [ [ [
] ] ] ] ] ]
[ ]
[ [ [ [ [
] ] ] ] ]
] ] ] ] ] ]
[ ]
[ [ [ [ [
] ] ] ] ]
[ ]
[ ]
[ ] [ ] [ ]
[ ] [ ] [ ]
[ ]
[ ]
[l
[1
interviewer blinded as to c h a r t status conducted a eomplaint-specific interview using a s t a n d a r d formatted questionnaire to ensure uniformity of patient interviews. Patients included in the follow-up portion of the study were contacted at specific intervals after their initial ED visit depending on the nature of their chief complaint. Nonpregnant women with complaints of abdominal pain or bleeding were called within seven days, whereas pregnant women with those complaints were contacted within two to three days. Patients with complaints of vaginal discharge only were contacted within seven days, or within 14 days if their diagnosis was candida vaginitis alone. Patients with initial complaints of abdominal pain were asked if they were still having pain (if so, whether it was the same, better, • Criterion Weight Possible Diagnoses Pelvic inflammatory disease/tube-ovarian 3 abscess Consult if fever more than 100 F, WBC more than 12,000, and there is marked tenderness on pelvic examination and peritoneal signs Appendicitis -- surgical consult Torsion of adnexa/tube -- gynecologic consult Rule-out endometfiosis -- gynecologic consult Ovarian cysts -- gynecologic consult and/or ultrasound in ED Ruptured cyst -- gynecologic consult Fibroids -- no consult; only gynecologic follow-up Gastroenteritis Abdominal pain of undetermined etiology Pyelonephritis Kidney stone Cystitis Treatment and Medication Pelvic inflammatory disease -- cephtriaxone and seven- to ten-day course of tetracycline/ vibriomycin (or spectinomycin for allergic patients) or 500 mg erythromycin Discharge Instructions For abdominal pain of undetermined etiology return within 24 hours -- All others make an appointment with private physician or gynecologic clinic within two days Return to the ED if symptoms persist, or return immediately if develop vaginal bleeding, fainting/lightheadedness, or fever Follow-up Call within seven days. Ask: Does patient still have pain? Has patient developed any other symptoms, including fever, bleeding, or fainting or lightheadedness? Has patient made a follow-up appointment? If not, why not?
Yes
No
[]
[]
[]
[ ]
[ ]
[]
536/73
PREFORMATTED CHARTS
Humphreys et al
or worse); if they were having other symptoms such as fever, bleeding, fainting, or lightheadedness; and if they were feeling completely better, somewhat better, slightly improved, no different, or worse. Patients with bleeding were asked the same questions as well as if they were still bleeding and if there was any change in the amount of bleeding. Patients with vaginal discharge were asked if they were still having a discharge or additional symptoms such as itching, burning, odor, pain, or fever and if they were feeling completely better, somewhat better, slightly improved, no different, or worse.
Patients also were asked to provide discharge educational information (ie, to recall what the physician said their diagnosis was, what medications they were taking, what was the dosing interval, and what were the follow-up instructions they had received). P a t i e n t s were asked to rate the quality of the medical care thcy reecivcd and the quality of the physician as excellent, good, fair, or poor. Demographic information such as birthdate, race, access to longitudinal care, and highest educational level achicvcd also was obtained at the time of the interview. Patients were assigned an outcome score based on the persistence of their complaints at the time of the follow-up interview. The most points were awarded for responses indicating resolution of specific complaints. Each patient's personal outcome score was converted to a percentage of total possible score. An outcome score of 100% was given to a patient feeling completely well with resolution of all symptoms initially present. Total chart score and the components (history, l a b o r a t o r y and physical examination, and patient's outcome score) were analyzed by two-way analysis of variance grouping on chief complaint and chart type. All scores were r e p o r t e d as a percentage of total possible score. The association between follow-up questionnaire items and c h a r t type was assessed by a two-tailed Pearson Z2 test or Fisher's exact test when a 2 × 2 contingency table had a cell with a frequency of 5 or less. Values of P < .05 were considered statistically significant.
Table 1.
Patient diagnoses No. of Patients
%
Abdominal pain of unknown etiology
22
13,8
Dysfunctional uterine bleeding
13
8.2
Physician Chart Diagnosis
Gastroenteritis
4
2.5
Intrauterine pregnancy
11
6.9
Pelvic inflammatory disease
49
25.2
Rule-out eotopic pregnancy
4
2.5
All statistical analyses were p e r f o r m e d using SAS PC software (SAS, Cary, North Carolina).
RESULTS One h u n d r e d fifty-nine charts (99 programmed and 60 blank) were evaluated. An additional 30 b l a n k charts were discarded because of illegibility, so a total of 30 b l a n k charts were studied. An estimated 15% to 20% of charts could not be retrieved for inclusion in the study because of admission of patients. Mean patient age was 28 + 10 years. There was no difference between the groups of patients with b l a n k and programmed charts with respect to chief complaint, age, race, or education. Eighty-one charts (50.9%) were those of nonpregnan t women with abdominal pain; 31 (19.5%) and 18 (11.3%) charts were those of nonpregnant women with bleeding and vaginal discharge, respectively; and 15 (9.4%) and 14 (8.8%) charts were those of pregnant women with pain and bleeding, respectively. The spectrum of patient diagnoses made in the ED is listed (Table 1). Sixteen charts (10%) were documented p r i m a r i l y by attending physicians, 41% (65) by postgraduate y e a r 3 residents, 29% (46) by postgraduate year 2 residents, and 19% (31) by interns, externs, or nurse practitioners. Two-way analysis of variance was used to examine documentation scores of the completed charts as well as components: history, l a b o r a t o r y studies, and physical examination as a function of chart type and complaint category. Regardless of complaint category, documentation was more complete on p r o g r a m m e d charts (81.1%) compared with blank charts (71.0%) (F = 21.5; df= 1,149; P = .0001). The greatest difference in c h a r t documentation scores was for nonpregnant women with abdominal pain (14% greater overall score in p r o g r a m m e d charts, P = .0001) (Table 2). The smallest difference was in charts of pregnant women with vaginal bleeding (6.3%, P = .28). Complaint category also was a significant factor in the observed differences in completeness of documentation. Charts of patients with potentially more serious complaints (eg, pregnant women with bleeding) tended to be associated with more complete documentation than did charts of patients with less severe complaints (eg, nonpregnant women with vaginal discharge; F = 6.41; df= 4,149; P = .0001). Because the differ•
Table 2.
Documentation score by chart type and complaint category Programmed Chart Complaint Category
Blank Chert
Mean Score
No.
Mean Score
No,
P ,0001
Threatened abortion
11
6.9
Not pregnant: pain
88.0
47
74.0
34
Urinary tract infection/pyelonephritis
13
8.2
Pregnant: pain
84.2
11
74.6
4
.1
Vaginitis
23
14.5
Not pregnant: bleeding
81.4
19
71.7
12
.01
Other
5
3,1
Pregnant: bleeding
80.0
9
73.7
5
.3
Not specified
6
3.8
Vaginal discharge
71.9
13
61.2
5
.05
74/537
ANNALS OF EMERGENCY MEDIClNE
21:5
MAY1992
PREFORMATTED CHARTS
Humphreys et al
ences observed between pregnant and n o n p r e g n a n t patients with the same complaints (pain or bleeding) were not significantly different from each other, the charts of these two patient groups were collapsed into categories based on chief complaint alone for further analyses. When the individual chart components, ie, history, physical examination, and laboratory studies, were examined, programmed charts significantly improved documentation of each component. Documentation of the patient history portion of the chart was more complete on programmed charts (74.8%) compared with b l a n k charts (60.1%) (F = 23.5; df= 1,153; P = .0001), with the largest improvement (16.6%) occurring in programmed charts of patients with abdominal pain (pain: mean score, 76.4; programmed chart mean score and number, 84.7 and 58; b l a n k chart mean score and n u m b e r , 68.1 and 38; P = .0001; bleeding: mean score, 67.5; programmed chart mean score and number, 74.3 and 28; b l a n k chart mean score and number, 60.7 and 17; P = .002; vaginal discharge: mean score, 58.4; programmed chart mean score and number, 65.4 and 13; b l a n k chart mean score and number, 51.4 and 5; P = .06; all complaints: programmed chart mean score and number, 74.8 and 99; b l a n k chart mean score and number, 60.1 and 60; P = .0001). As with overall score, patients with more severe complaints had the history portion of their charts better documented regardless of chart type (F = 13.3; df= 2,153; P = .0001). Physical examinations also were more completely documented when programmed charts were used (F = 10.68; df= 1,153; P = .0013). However, when examined by chief complaint, only patients with more severe complaints showed a statistically significant improvement (pain, P = .0001; bleeding, P = .035) in chart documentation when programmed charts were used (pain: mean score, 81.6; programmed chart mean score and number, 91.4 a n d 58; b l a n k chart mean score and number, 71.7 and 38; P = .0001; bleeding: mean score, 80.7; programmed chart mean score and number, 86.1 and 28; b l a n k chart mean score and number, 75.3 and 17; P = .035; vaginal discharge: mean score, 86.4; programmed chart mean score and number, 88.5 and 13; b l a n k chart mean score and number, 84.3 and 5; P = .6; all complaints: programmed chart mean score and number, 88.7 and 99; b l a n k chart mean score and number, 77.1 and 60; P = .001). Programmed charts of patients with a chief complaint of abdominal pain or bleeding had a 19.8% and 10.8% improvement in chart documentation, respectively, whereas programmed charts of patients with a chief complaint of vaginal discharge only had a 4.2% improvement. Documentation of laboratory studies also was more complete on programmed charts (83.3%) compared with b l a n k charts (73.3%) (F = 4.6; df= 1,153; P = .03) (pain: mean score, 84.0; programmed chart mean score and number, 88.4 and 58; b l a n k chart mean score and n u m b e r , 79.6 and 38; P = .05; bleeding: mean score, 91.7; programmed chart mean score and number, 90.8 a n d 28; b l a n k chart mean
MAY1992
21:5 ANNALS OF EMERGENCY MEDICINE
score and number, 92.6 and 17; P = .8; vaginal discharge: mean score, 59.4; programmed chart mean score and number, 70.8 and 13; b l a n k chart mean score and number, 48.0 and 5; P = .05; all complaints: programmed chart mean score and number, 83.3 and 99; b l a n k chart mean score and number, 73.4 and 60; P = .03). When examined by chief complaint, programmed chart documentation of laboratory studies was significantly better than b l a n k chart documentation for patients with a chief complaint of pain (P = .05) and vaginal discharge (P = .05) but not bleeding (P = .8). Regardless of chart type, laboratory study documentation differed by chief complaint (F = 12.1; df= 2,153; P = .0001). Charts of patients with a chief complaint of bleeding (91.7%) were more complete with regard to laboratory studies compared with charts of patients with pain (84.0%) and vaginal discharge (59.4%). Documentation of charts varied with the level of physician training. Postgraduate year 3 residents had the highest documentation scores for both programmed and b l a n k charts followed by interns/externs and postgraduate year 2 residents (attending physician: mean score, 62.8; programmed chart mean score and number, 72.7 and 9; b l a n k chart mean score and number, 52.9 and 7; P = .0002; postgraduate year 3 residents: mean score, 81.5; programmed chart mean score and number, 82.3 and 32; b l a n k chart mean score and number, 80.7 and 14; P = .7; postgraduate year 2 residents: mean score, 74.8; programmed chart mean score and number, 80.0 and 39; b l a n k chart mean score and number, 69.7 and 26; P = .008; interns/externs: mean score, 78.7; programmed chart mean score and number, 81.5 and 19; b l a n k chart mean score and number, 75.9 and 12; P = .2). Attending physicians had the lowest documentation scores for both chart types. Attending physicians had the greatest significant difference in documentation performance with programmed charts (19.8%), postgraduate year 2 residents had 10.2% better documentation of programmed charts, and no significant difference was observed in the documentation practices of postgraduate year 3 residents or interns/externs. One h u n d r e d eighteen patients were Contacted successfully for follow-up interviews after their initial ED visit. When percent outcome scores based on the persistence of symptoms were examined, outcome scores were correlated with severity of complaint (F = 5.3; df= 2,112; P = .006) b u t not with chart type (F = 0; df= 1,112; P = .9). Patients treated for vaginal discharge had the highest average outcome score (90.8%) compared with patients treated for pain (81.9%) or bleeding (73.2%). When patients were asked to evaluate the quality of care that they received, there was no statistically significant difference i n the ratings given by patients whose physicians had used programmed charts compared with those whose physicians had not. Likewise, there was no difference in patients' rating of the professional competence of the physicians responsible for their care in the ED. •
538/75
PREFORMATTED
CHARTS
Humphreys et al
During the follow-up interview, patients were asked to provide the diagnosis given to them in the ED. When the diagnosis provided by the patients was c o m p a r e d with that written in the c h a r t by the physician, there was a t r e n d toward greater agreement when p r o g r a m m e d charts (84.5%) were used than with b l a n k charts (72.3%). This difference was not statistically significant (X2 = 1.9, P = .2). However, when p r o g r a m m e d charts were used, 86.5% of these patients (71) thought that their problem was explained satisfactorily c o m p a r e d with 69.4% of patients (47) for whom b l a n k charts were used (X2 = 5.3, P = .02). When patients were asked about their follow-up plans and compliance with medications that had been p r e s c r i b e d , similar trends were observed. Seventy-six percent of the patients in the p r o g r a m m e d c h a r t group (50) h a d made follow-up appointments when instructed to do so c o m p a r e d with 55% of patients in the b l a n k - c h a r t group. Ninety-four percent of patients (78) to whom medications h a d been prescribed r e p o r t e d high rates of compliance. There was a trend toward b e t t e r compliance in patients whose physicians had used p r o g r a m m e d charts (98% vs 87%). Both findings were trends and lacked statistical significance (P = .08 and P = .07, respectively).
DISCUSSION The data generated by this study support the hypothesis that p r o g r a m m e d charts are associated with more complete physician documentation of patient encounters and refute the null hypothesis that there is no difference in completeness of documentation when blank and p r o g r a m m e d charts are used. Patient history documentation a p p e a r e d to benefit most from the use of programmed charts, whereas laboratory studies seemed to benefit the least. Completeness of documentation also was positively associated with the severity of the complaint. This finding is in agreement with that of other studies examining the c h a r t documentation practices of medical residents. 5 The ability of programmed charts to improve documentation by physicians has several i m p o r t a n t implications. Programmed charts may be useful in improving communication between care providers. They also may aid in protecting EDs from malpractice litigation arising from insufficient documentation. A previously published review of ED malpractice claims established that the most common record deficiency was failure to document essential historical and physical examination data. 6 P r o g r a m m e d charts also may help improve efficiency of data recording. Additional studies are needed to demonstrate the utility of p r o g r a m m e d charts in saving time spent charting and in reducing malpractice claims. The use of programmed charts had no significant effect on patient outcome p a r a m e t e r s as measured by outcome score at the time of the follow-up interview. This is consistent with
76/539
the notion that medical records are not an accurate reflection of the quality of patient care received or extent of patient workup. As expected, the severity of the complaint was most useful in predicting outcome. Patients with more severe complaints had lower outcome scores. However, the time elapsed from initial ED visit to follow-up interview may be a confounder (patients with more serious complaints were contacted sooner and therefore had less resolution of their symptoms). The lack of correlation between completeness of documentation and patient outcome is an important finding that is consistent with previous studies conducted with different patient populations. 1 A study of hospital inpatients with appendicitis or myocardial infarction found that the completeness of recording was unrelated to outcome. 4 This finding c o r r o b o r a t e d subsequently for additional inpatients with pneumonia and myocardial infarction. 3 Likewise, the use of problem-oriented medical records in inpatients with anemia resulted in no improvement in clinical outcome. 2 In the ED setting, the quality of medical records was not found to be related to the health status outcome of patients with acute bronchial asthma. 1 Our study also suggests that chart documentation is not an accurate reflection of the quality of care given. However, a larger sample size of patients successfully contacted for follow-up is desirable before finalizing these conclusions. P e r h a p s other p a r a m e t e r s must be considered to more precisely assess patient care as it relates to health care outcome. F u r t h e r support for this belief comes from the follow-up surveys of the patients themselves. There was no difference in the patients' own assessment of the care they received relative to degree of documentation. However, there was an observable, although not statistically significant, difference in patient recollection of their diagnosis when p r o g r a m m e d charts were used. Patients tended to exhibit a greater accuracy in reporting their diagnosis and greater satisfaction with their physician's explanation of that diagnosis when p r o g r a m m e d charts were used. One possible explanation of this phenomenon is that p r o g r a m m e d charts reduced documentation time and permitted more attention to the details of patient care, including communication skills. A previous study of residents' interaction with patients noted the greatest deficiencies were in the a r e a of patient education. 5 This aspect of p r o g r a m m e d charts deserves exploration in future studies. It is possible that better patient comprehension of their diagnosis may be the m a j o r factor to improved compliance and follow-up as r e p o r t e d by patients when programmed charts were used. It is surprising that p r o g r a m m e d charts increased documentation for attending physicians and postgraduate y e a r 2 residents, whereas postgraduate y e a r 3 residents and interns/externs showed little difference in their documentation practices. One would have expected the greatest improvement in physicians with the lowest level of training. This finding may challenge the utility of p r o g r a m m e d charts as a teaching tool. P e r h a p s less-experienced housestaff
ANNALS OF EMERGENCY MEDIOINE
21:5
MAY1992
PREFORMATTED
CHARTS
Humphreys et al
tend to maintain a similar level of compulsiveness in chart documentation regardless of chart type. Attending physicians who are confident of their diagnoses may spend less time documenting charts thoroughly unless presented with specific preformatted patient records. One limitation of this study is the possible bias arising from the inability to blind the chart reviewer to chart type. Another limitation of this study relates to its restriction to a specific group of patients, ie, women of child-bearing age with obstetric and gynecologic complaints. Programmed charts may have limited utility with higher acuity patients who have complaints such as chest pain, trauma, or toxic ingestion that may compel the physician to document a more comprehensive multisystem clinical base. One may also argue that if programmed charts have no direct effect on patient health outcome, then their usefulness is limited. We believe that they represent a useful option for the ED physician for improving the documentation of patient care and thereby decreasing medicolegal vulnerability and fostering interdepartmental communication. During the data-gathering process, approximately one third of blank charts had to be discarded because of illegible handwriting. Although programmed charts will not improve the handwriting clarity of physicians, the check-off format and prompts of programmed charts are less susceptible to illegibility. The use of preformatted charts may expedite and facilitate the documentation process, thereby allowing more time for patient education. Additional studies are required to demonstrate this. Programmed charts also may facilitate research that uses chart reviews because these charts require uniform responses and improve completeness of documentation.
CONCLUSION
Programmed charts resulted in more complete documentation of ED records for a selected group of patients. The greatest improvements were in the recording of the history and physical examination. There was no detectable correlation between the use of programmed charts and patient health status outcomes, patient satisfaction with care, or patient perception of physician quality. There was a positive correlation with the use of preformatted charts and the patients' understanding of their discharge diagnosis, compliance with medications, and follow-up instructions. •
REFERENCES 1. Murphy JG, Jacobsen S: Assessing the quality of emergency care: The medical record versus patients' outcomes. Ann EmergMed 1984:13:158-165. 2. Switz D: The problem-oriented medical record. Arch/ntem Med 1976;136:1119-1123. 3. Sanazaro PJ, Worth RM: Concurrent quality assurance in hospital care. N EnglJ Med1974;298:1171-1177. 4. Fessel W J, Van Brunt EF: Assessing quality of care from the medical record. N Eng/J Med 1972;286:134-138. 5. Maran WT, Wiser TH, Nanda J, et al: Measuring medical residents chart documentation practices. J Med Educ 1988;63:859-865. 6. Traulhein J: Malpractice in the emergency department-- A review of 200 cases. Ann EmergMed 1984;13:709-711.
Address for reprints: Sheldon Jacobsen, MD, Emergency Services Department, Ground Floor Silverstein, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, Pennsylvania 19104.
i
L ii ! MAY 1992
21:5
ANNALS OF EMERGENCY MEDICINE
540177
documentation; medical records
Preformatted ChartsImproveDocumentationin the EmergencyDepartment From the University of Pennsylvania School of Medicine;* the Emergency
Tanya Humphreys,MD* Frances S Shofer, PhD*" Sheldon Jacobson, MD*§
Servicest and Obstetrics and
Christos Coutifaris, MD, PhD~
Gynecologyt Departments,
Annette Stemhagen, DrPHt§
Study objectives: Todetermine if the use of programmed charts with complaint-specific entry criteria results in improved documentation of patient encounters and better clinical outcome. Design: Prospective study.
Hospital of the University of
Setting: Emergency department of an urban university hospital.
Pennsylvania; and the Section
Type of participants: Female patients presenting to the emergency
of General Internal Medicine,
department with gynecologic complaints of abdominal pain, bleeding, or vaginal discharge.
Department of Medicine, University of Pennsylvania, Philadelphia.§
Interventions:
Received for publication November 26, 1990. Revision
Measurements:
received September 23, 1991. Accepted for publication January 28, 1992.
Presented at the Societyfor Academic Emergency Medicine Annual Meeting in Minneapolis,
Minnesota, May 1990.
Programmed and blank charts were provided randemly for physicians in the ED.
Chart scores based on documentation criteria for patient history, physical examination, laboratory studies, diagnosis, and discharge instructions and patient outcome scores of 0% to 100% based on the persistence of their complaints at the time of the followup interview. Main results: Overall documentation of history, physical examination, and laboratory studies was more complete on programmed charts than on blank charts (81.1% vs 71%, P< .0001). The patient history portion of the charts was found to benefit the most from the use of programmed charts (74.8% vs 60.1%, P< .0001). Although programmed charts demonstrated better documentation, there was no statistically significant correlation with patient outcome parameters or with patient satisfaction with the quality of medical care. However, more patients whose physicians used programmed charts were satisfied with their physicians' explanations of their problem (X~ = 5.2, P< ,02). Conclusion: Programmed charts improve documentation by facilitation of the documentation process and allow more time for patientphysician interaction. Quality of documentation alone, however, is not a reliable indicator of patient outcome or of the quality of care received. [Humphreys T, Shofer FS, Jacobson S, Coutifaris C, Stemhagen A: Preformatted Charts improve documentation in the emergency department. Ann EmergMad May 1992;21:534-540.]
MAY 1992 21:5 ANNALS OF EMER6ENCY MEDiCiNE
53 4 / 7 1
PREFORMATTED
CHARTS
Humphreys et al
INTRODUCTION
Preformatted problem-oriented patient encounter charts (programmed charts) have been used sporadically by emergency physicians as a means of simultaneously increasing the accuracy of documentation and reducing the amount of time spent documenting patient care. These charts, which provide a complete list of important documentation criteria for different medical complaints, also may be a valuable teaching tool for housestaff training. The relationship between the quality of the medical records and patient outcome remains controversial. There is no conclusive evidence that the completeness of information recording in medical records reflects the quality of medical care given. Many studies, in fact, suggest that medical records alone are not a reliable indicator of patient care or health outcome, x-a However, theissue of accurate documentation is an important one, as medical records are used frequently by third-party payers to validate claims and by hospital administrators and lawyers to evaluate physician performance for quality assurance audits and malpractice
Figure 1. Programmed chart
]
~ ' ~
•
U
•
EMERGENCYSERVICES MEDICAL RIPORT
'
F
I"
s.u,s.:
~G
(_ E ~ R C e N C V S~RV,CeS O E P A " ~ E N T
72/535
F
/ ........
GYN/ 08 UTI
I ....
P,~.~:
o ua,s,m,
o ~
suits. More important, medical records are a vital means of communication between patient care providers. Completeness of documentation represents a special challenge for emergency physicians who work in high-volume, high-acuity settings with competing demands of patient care and administrative responsibilities. We sought to establish the ability of programmed charts to improve documentation in the El) setting and to evaluate the effect, if any, of preformatted charts on the clinical outcomes of patients.
MATERIALS
AND
METHODS
Programmed charts were designed by the attending ED staff at the Hospital of the University of Pennsylvania for recording obstetric and gynecologic evaluations in the El) (Figure 1). These charts are preformatted with specific prompts for information pertaining to patient history and physical examination. The study was conducted in the ED of the Hospital of the University of Pennsylvania, a large urban teaching hospital with 45,000 ED visits annually. Approximately 15% of these visits are for obstetric and gynecologic complaints; of these, 15% to 20% are admitted. Between February 1989 and August 1989, either standard blank charts or preformatted programmed charts were provided on alternate weeks by ED secretaries to internal medicine housestaff and emergency attending physicians for the documentation of patient encounters. Approximately 5% of completed programmed and blank charts of patients between the ages of 15 and 45 years seeking care in the ED for obstetric or gynecologic problems were selected randomly for evaluation of completeness. Relevant charts were identified and retrieved through the use of the El) daily log by a trained volunteer. Because of the medical records practices in our hospital, charts of patients who were admitted were not available for inclusion in this study. Charts that were illegible also were not included for evaluation. All patients were blinded as to chart type used. A panel of three obstetrician/gynecologists and one emergency physician developed a set of documentation criteria for patient history, physical examination, laboratory studies, diagnosis, and discharge instructions. Separate documentation criteria were established by group consensus for five categories of patient complaints frequently encountered in the El): pregnant and nonpregnant women with abdominal pain, pregnant and nonpregnant women with vaginal bleeding, and nonpregnant women with vaginal discharge. Then, each individual documentation criterion was assigned a weighted value (1 through 3) based on its relative importance with respect to the specific complaint category as determined by the panel. Examples of documentation criteria for nonpregnant women with abdominal pain included specific historical information, physical examination findings, and laboratory criteria (Figure 2). Similar •
ANNALS 0FEMER6ENCYMEDICtNE
21:5
MAY1992
PREFORMATTED CHARTS f l u m p h r e y s et al
criteria were applied to charts of the other categories of chief complaints. Each completed c h a r t (programmed and blank) was reviewed by a single reviewer (not p a r t of the original panel) using a s t a n d a r d evaluation sheet formatted for each complaint category, and a c h a r t score was calculated based on the presence or absence of criteria deemed necessary for complete documentation. Then, the percentage of the total possible score was calculated for each c h a r t to allow greater ease of comparison among categories. Attempts were made to contact by telephone all patients whose charts had been collected for evaluation. A trained Figure2. Documentation criteria for nonpregnant women with abdominal pain
Criterion Weight 3 3 2
2 3 3 1 1 1 3 2 3 3
2 3 3 3 3
History Location of pain Duration of pain Radiation Where did pain start? Has it moved? Bowel symptoms or gastrointestinal complaints (nausea, vomiting, diarrhea) Last menstrual period History of fever Other medical problems, including surgical history Medications Birth control pill use Allergies Vaginal discharge Sexual activity Reproductive history (gravity, parity, history of abortions, past pregnancy problems, pelvic inflammatory disease history) Physical Examination Cardiopulmonary Pelvic Rectal Flanks (costovertebra[ angle tenderness) Abdomen Laboratory Pregnancy test -- negative Urinalysis (microanalysis or sedimentation and dipstick) CBC Cultures If there is a mucopurulent discharge Cervical Vaginal wet mount (saline) Consultation and Diagnosis If consult needed and is called (no extra credit if mere than one consult is called)
MAY1992
21:5 ANNALS 0F EMERGENCY MEDICINE
Yes [ ] [ ]
No [ ] [ ]
[ ] [ ]
[ ] [ ]
[ ] [ ] [ ]
[ ] [ ] [ ]
[ [ [ [ [ [
[ [ [ [ [ [
] ] ] ] ] ]
[ ]
[ [ [ [ [
] ] ] ] ]
] ] ] ] ] ]
[ ]
[ [ [ [ [
] ] ] ] ]
[ ]
[ ]
[ ] [ ] [ ]
[ ] [ ] [ ]
[ ]
[ ]
[l
[1
interviewer blinded as to c h a r t status conducted a eomplaint-specific interview using a s t a n d a r d formatted questionnaire to ensure uniformity of patient interviews. Patients included in the follow-up portion of the study were contacted at specific intervals after their initial ED visit depending on the nature of their chief complaint. Nonpregnant women with complaints of abdominal pain or bleeding were called within seven days, whereas pregnant women with those complaints were contacted within two to three days. Patients with complaints of vaginal discharge only were contacted within seven days, or within 14 days if their diagnosis was candida vaginitis alone. Patients with initial complaints of abdominal pain were asked if they were still having pain (if so, whether it was the same, better, • Criterion Weight Possible Diagnoses Pelvic inflammatory disease/tube-ovarian 3 abscess Consult if fever more than 100 F, WBC more than 12,000, and there is marked tenderness on pelvic examination and peritoneal signs Appendicitis -- surgical consult Torsion of adnexa/tube -- gynecologic consult Rule-out endometfiosis -- gynecologic consult Ovarian cysts -- gynecologic consult and/or ultrasound in ED Ruptured cyst -- gynecologic consult Fibroids -- no consult; only gynecologic follow-up Gastroenteritis Abdominal pain of undetermined etiology Pyelonephritis Kidney stone Cystitis Treatment and Medication Pelvic inflammatory disease -- cephtriaxone and seven- to ten-day course of tetracycline/ vibriomycin (or spectinomycin for allergic patients) or 500 mg erythromycin Discharge Instructions For abdominal pain of undetermined etiology return within 24 hours -- All others make an appointment with private physician or gynecologic clinic within two days Return to the ED if symptoms persist, or return immediately if develop vaginal bleeding, fainting/lightheadedness, or fever Follow-up Call within seven days. Ask: Does patient still have pain? Has patient developed any other symptoms, including fever, bleeding, or fainting or lightheadedness? Has patient made a follow-up appointment? If not, why not?
Yes
No
[]
[]
[]
[ ]
[ ]
[]
536/73
PREFORMATTED CHARTS
Humphreys et al
or worse); if they were having other symptoms such as fever, bleeding, fainting, or lightheadedness; and if they were feeling completely better, somewhat better, slightly improved, no different, or worse. Patients with bleeding were asked the same questions as well as if they were still bleeding and if there was any change in the amount of bleeding. Patients with vaginal discharge were asked if they were still having a discharge or additional symptoms such as itching, burning, odor, pain, or fever and if they were feeling completely better, somewhat better, slightly improved, no different, or worse.
Patients also were asked to provide discharge educational information (ie, to recall what the physician said their diagnosis was, what medications they were taking, what was the dosing interval, and what were the follow-up instructions they had received). P a t i e n t s were asked to rate the quality of the medical care thcy reecivcd and the quality of the physician as excellent, good, fair, or poor. Demographic information such as birthdate, race, access to longitudinal care, and highest educational level achicvcd also was obtained at the time of the interview. Patients were assigned an outcome score based on the persistence of their complaints at the time of the follow-up interview. The most points were awarded for responses indicating resolution of specific complaints. Each patient's personal outcome score was converted to a percentage of total possible score. An outcome score of 100% was given to a patient feeling completely well with resolution of all symptoms initially present. Total chart score and the components (history, l a b o r a t o r y and physical examination, and patient's outcome score) were analyzed by two-way analysis of variance grouping on chief complaint and chart type. All scores were r e p o r t e d as a percentage of total possible score. The association between follow-up questionnaire items and c h a r t type was assessed by a two-tailed Pearson Z2 test or Fisher's exact test when a 2 × 2 contingency table had a cell with a frequency of 5 or less. Values of P < .05 were considered statistically significant.
Table 1.
Patient diagnoses No. of Patients
%
Abdominal pain of unknown etiology
22
13,8
Dysfunctional uterine bleeding
13
8.2
Physician Chart Diagnosis
Gastroenteritis
4
2.5
Intrauterine pregnancy
11
6.9
Pelvic inflammatory disease
49
25.2
Rule-out eotopic pregnancy
4
2.5
All statistical analyses were p e r f o r m e d using SAS PC software (SAS, Cary, North Carolina).
RESULTS One h u n d r e d fifty-nine charts (99 programmed and 60 blank) were evaluated. An additional 30 b l a n k charts were discarded because of illegibility, so a total of 30 b l a n k charts were studied. An estimated 15% to 20% of charts could not be retrieved for inclusion in the study because of admission of patients. Mean patient age was 28 + 10 years. There was no difference between the groups of patients with b l a n k and programmed charts with respect to chief complaint, age, race, or education. Eighty-one charts (50.9%) were those of nonpregnan t women with abdominal pain; 31 (19.5%) and 18 (11.3%) charts were those of nonpregnant women with bleeding and vaginal discharge, respectively; and 15 (9.4%) and 14 (8.8%) charts were those of pregnant women with pain and bleeding, respectively. The spectrum of patient diagnoses made in the ED is listed (Table 1). Sixteen charts (10%) were documented p r i m a r i l y by attending physicians, 41% (65) by postgraduate y e a r 3 residents, 29% (46) by postgraduate year 2 residents, and 19% (31) by interns, externs, or nurse practitioners. Two-way analysis of variance was used to examine documentation scores of the completed charts as well as components: history, l a b o r a t o r y studies, and physical examination as a function of chart type and complaint category. Regardless of complaint category, documentation was more complete on p r o g r a m m e d charts (81.1%) compared with blank charts (71.0%) (F = 21.5; df= 1,149; P = .0001). The greatest difference in c h a r t documentation scores was for nonpregnant women with abdominal pain (14% greater overall score in p r o g r a m m e d charts, P = .0001) (Table 2). The smallest difference was in charts of pregnant women with vaginal bleeding (6.3%, P = .28). Complaint category also was a significant factor in the observed differences in completeness of documentation. Charts of patients with potentially more serious complaints (eg, pregnant women with bleeding) tended to be associated with more complete documentation than did charts of patients with less severe complaints (eg, nonpregnant women with vaginal discharge; F = 6.41; df= 4,149; P = .0001). Because the differ•
Table 2.
Documentation score by chart type and complaint category Programmed Chart Complaint Category
Blank Chert
Mean Score
No.
Mean Score
No,
P ,0001
Threatened abortion
11
6.9
Not pregnant: pain
88.0
47
74.0
34
Urinary tract infection/pyelonephritis
13
8.2
Pregnant: pain
84.2
11
74.6
4
.1
Vaginitis
23
14.5
Not pregnant: bleeding
81.4
19
71.7
12
.01
Other
5
3,1
Pregnant: bleeding
80.0
9
73.7
5
.3
Not specified
6
3.8
Vaginal discharge
71.9
13
61.2
5
.05
74/537
ANNALS OF EMERGENCY MEDIClNE
21:5
MAY1992
PREFORMATTED CHARTS
Humphreys et al
ences observed between pregnant and n o n p r e g n a n t patients with the same complaints (pain or bleeding) were not significantly different from each other, the charts of these two patient groups were collapsed into categories based on chief complaint alone for further analyses. When the individual chart components, ie, history, physical examination, and laboratory studies, were examined, programmed charts significantly improved documentation of each component. Documentation of the patient history portion of the chart was more complete on programmed charts (74.8%) compared with b l a n k charts (60.1%) (F = 23.5; df= 1,153; P = .0001), with the largest improvement (16.6%) occurring in programmed charts of patients with abdominal pain (pain: mean score, 76.4; programmed chart mean score and number, 84.7 and 58; b l a n k chart mean score and n u m b e r , 68.1 and 38; P = .0001; bleeding: mean score, 67.5; programmed chart mean score and number, 74.3 and 28; b l a n k chart mean score and number, 60.7 and 17; P = .002; vaginal discharge: mean score, 58.4; programmed chart mean score and number, 65.4 and 13; b l a n k chart mean score and number, 51.4 and 5; P = .06; all complaints: programmed chart mean score and number, 74.8 and 99; b l a n k chart mean score and number, 60.1 and 60; P = .0001). As with overall score, patients with more severe complaints had the history portion of their charts better documented regardless of chart type (F = 13.3; df= 2,153; P = .0001). Physical examinations also were more completely documented when programmed charts were used (F = 10.68; df= 1,153; P = .0013). However, when examined by chief complaint, only patients with more severe complaints showed a statistically significant improvement (pain, P = .0001; bleeding, P = .035) in chart documentation when programmed charts were used (pain: mean score, 81.6; programmed chart mean score and number, 91.4 a n d 58; b l a n k chart mean score and number, 71.7 and 38; P = .0001; bleeding: mean score, 80.7; programmed chart mean score and number, 86.1 and 28; b l a n k chart mean score and number, 75.3 and 17; P = .035; vaginal discharge: mean score, 86.4; programmed chart mean score and number, 88.5 and 13; b l a n k chart mean score and number, 84.3 and 5; P = .6; all complaints: programmed chart mean score and number, 88.7 and 99; b l a n k chart mean score and number, 77.1 and 60; P = .001). Programmed charts of patients with a chief complaint of abdominal pain or bleeding had a 19.8% and 10.8% improvement in chart documentation, respectively, whereas programmed charts of patients with a chief complaint of vaginal discharge only had a 4.2% improvement. Documentation of laboratory studies also was more complete on programmed charts (83.3%) compared with b l a n k charts (73.3%) (F = 4.6; df= 1,153; P = .03) (pain: mean score, 84.0; programmed chart mean score and number, 88.4 and 58; b l a n k chart mean score and n u m b e r , 79.6 and 38; P = .05; bleeding: mean score, 91.7; programmed chart mean score and number, 90.8 a n d 28; b l a n k chart mean
MAY1992
21:5 ANNALS OF EMERGENCY MEDICINE
score and number, 92.6 and 17; P = .8; vaginal discharge: mean score, 59.4; programmed chart mean score and number, 70.8 and 13; b l a n k chart mean score and number, 48.0 and 5; P = .05; all complaints: programmed chart mean score and number, 83.3 and 99; b l a n k chart mean score and number, 73.4 and 60; P = .03). When examined by chief complaint, programmed chart documentation of laboratory studies was significantly better than b l a n k chart documentation for patients with a chief complaint of pain (P = .05) and vaginal discharge (P = .05) but not bleeding (P = .8). Regardless of chart type, laboratory study documentation differed by chief complaint (F = 12.1; df= 2,153; P = .0001). Charts of patients with a chief complaint of bleeding (91.7%) were more complete with regard to laboratory studies compared with charts of patients with pain (84.0%) and vaginal discharge (59.4%). Documentation of charts varied with the level of physician training. Postgraduate year 3 residents had the highest documentation scores for both programmed and b l a n k charts followed by interns/externs and postgraduate year 2 residents (attending physician: mean score, 62.8; programmed chart mean score and number, 72.7 and 9; b l a n k chart mean score and number, 52.9 and 7; P = .0002; postgraduate year 3 residents: mean score, 81.5; programmed chart mean score and number, 82.3 and 32; b l a n k chart mean score and number, 80.7 and 14; P = .7; postgraduate year 2 residents: mean score, 74.8; programmed chart mean score and number, 80.0 and 39; b l a n k chart mean score and number, 69.7 and 26; P = .008; interns/externs: mean score, 78.7; programmed chart mean score and number, 81.5 and 19; b l a n k chart mean score and number, 75.9 and 12; P = .2). Attending physicians had the lowest documentation scores for both chart types. Attending physicians had the greatest significant difference in documentation performance with programmed charts (19.8%), postgraduate year 2 residents had 10.2% better documentation of programmed charts, and no significant difference was observed in the documentation practices of postgraduate year 3 residents or interns/externs. One h u n d r e d eighteen patients were Contacted successfully for follow-up interviews after their initial ED visit. When percent outcome scores based on the persistence of symptoms were examined, outcome scores were correlated with severity of complaint (F = 5.3; df= 2,112; P = .006) b u t not with chart type (F = 0; df= 1,112; P = .9). Patients treated for vaginal discharge had the highest average outcome score (90.8%) compared with patients treated for pain (81.9%) or bleeding (73.2%). When patients were asked to evaluate the quality of care that they received, there was no statistically significant difference i n the ratings given by patients whose physicians had used programmed charts compared with those whose physicians had not. Likewise, there was no difference in patients' rating of the professional competence of the physicians responsible for their care in the ED. •
538/75
PREFORMATTED
CHARTS
Humphreys et al
During the follow-up interview, patients were asked to provide the diagnosis given to them in the ED. When the diagnosis provided by the patients was c o m p a r e d with that written in the c h a r t by the physician, there was a t r e n d toward greater agreement when p r o g r a m m e d charts (84.5%) were used than with b l a n k charts (72.3%). This difference was not statistically significant (X2 = 1.9, P = .2). However, when p r o g r a m m e d charts were used, 86.5% of these patients (71) thought that their problem was explained satisfactorily c o m p a r e d with 69.4% of patients (47) for whom b l a n k charts were used (X2 = 5.3, P = .02). When patients were asked about their follow-up plans and compliance with medications that had been p r e s c r i b e d , similar trends were observed. Seventy-six percent of the patients in the p r o g r a m m e d c h a r t group (50) h a d made follow-up appointments when instructed to do so c o m p a r e d with 55% of patients in the b l a n k - c h a r t group. Ninety-four percent of patients (78) to whom medications h a d been prescribed r e p o r t e d high rates of compliance. There was a trend toward b e t t e r compliance in patients whose physicians had used p r o g r a m m e d charts (98% vs 87%). Both findings were trends and lacked statistical significance (P = .08 and P = .07, respectively).
DISCUSSION The data generated by this study support the hypothesis that p r o g r a m m e d charts are associated with more complete physician documentation of patient encounters and refute the null hypothesis that there is no difference in completeness of documentation when blank and p r o g r a m m e d charts are used. Patient history documentation a p p e a r e d to benefit most from the use of programmed charts, whereas laboratory studies seemed to benefit the least. Completeness of documentation also was positively associated with the severity of the complaint. This finding is in agreement with that of other studies examining the c h a r t documentation practices of medical residents. 5 The ability of programmed charts to improve documentation by physicians has several i m p o r t a n t implications. Programmed charts may be useful in improving communication between care providers. They also may aid in protecting EDs from malpractice litigation arising from insufficient documentation. A previously published review of ED malpractice claims established that the most common record deficiency was failure to document essential historical and physical examination data. 6 P r o g r a m m e d charts also may help improve efficiency of data recording. Additional studies are needed to demonstrate the utility of p r o g r a m m e d charts in saving time spent charting and in reducing malpractice claims. The use of programmed charts had no significant effect on patient outcome p a r a m e t e r s as measured by outcome score at the time of the follow-up interview. This is consistent with
76/539
the notion that medical records are not an accurate reflection of the quality of patient care received or extent of patient workup. As expected, the severity of the complaint was most useful in predicting outcome. Patients with more severe complaints had lower outcome scores. However, the time elapsed from initial ED visit to follow-up interview may be a confounder (patients with more serious complaints were contacted sooner and therefore had less resolution of their symptoms). The lack of correlation between completeness of documentation and patient outcome is an important finding that is consistent with previous studies conducted with different patient populations. 1 A study of hospital inpatients with appendicitis or myocardial infarction found that the completeness of recording was unrelated to outcome. 4 This finding c o r r o b o r a t e d subsequently for additional inpatients with pneumonia and myocardial infarction. 3 Likewise, the use of problem-oriented medical records in inpatients with anemia resulted in no improvement in clinical outcome. 2 In the ED setting, the quality of medical records was not found to be related to the health status outcome of patients with acute bronchial asthma. 1 Our study also suggests that chart documentation is not an accurate reflection of the quality of care given. However, a larger sample size of patients successfully contacted for follow-up is desirable before finalizing these conclusions. P e r h a p s other p a r a m e t e r s must be considered to more precisely assess patient care as it relates to health care outcome. F u r t h e r support for this belief comes from the follow-up surveys of the patients themselves. There was no difference in the patients' own assessment of the care they received relative to degree of documentation. However, there was an observable, although not statistically significant, difference in patient recollection of their diagnosis when p r o g r a m m e d charts were used. Patients tended to exhibit a greater accuracy in reporting their diagnosis and greater satisfaction with their physician's explanation of that diagnosis when p r o g r a m m e d charts were used. One possible explanation of this phenomenon is that p r o g r a m m e d charts reduced documentation time and permitted more attention to the details of patient care, including communication skills. A previous study of residents' interaction with patients noted the greatest deficiencies were in the a r e a of patient education. 5 This aspect of p r o g r a m m e d charts deserves exploration in future studies. It is possible that better patient comprehension of their diagnosis may be the m a j o r factor to improved compliance and follow-up as r e p o r t e d by patients when programmed charts were used. It is surprising that p r o g r a m m e d charts increased documentation for attending physicians and postgraduate y e a r 2 residents, whereas postgraduate y e a r 3 residents and interns/externs showed little difference in their documentation practices. One would have expected the greatest improvement in physicians with the lowest level of training. This finding may challenge the utility of p r o g r a m m e d charts as a teaching tool. P e r h a p s less-experienced housestaff
ANNALS OF EMERGENCY MEDIOINE
21:5
MAY1992
PREFORMATTED
CHARTS
Humphreys et al
tend to maintain a similar level of compulsiveness in chart documentation regardless of chart type. Attending physicians who are confident of their diagnoses may spend less time documenting charts thoroughly unless presented with specific preformatted patient records. One limitation of this study is the possible bias arising from the inability to blind the chart reviewer to chart type. Another limitation of this study relates to its restriction to a specific group of patients, ie, women of child-bearing age with obstetric and gynecologic complaints. Programmed charts may have limited utility with higher acuity patients who have complaints such as chest pain, trauma, or toxic ingestion that may compel the physician to document a more comprehensive multisystem clinical base. One may also argue that if programmed charts have no direct effect on patient health outcome, then their usefulness is limited. We believe that they represent a useful option for the ED physician for improving the documentation of patient care and thereby decreasing medicolegal vulnerability and fostering interdepartmental communication. During the data-gathering process, approximately one third of blank charts had to be discarded because of illegible handwriting. Although programmed charts will not improve the handwriting clarity of physicians, the check-off format and prompts of programmed charts are less susceptible to illegibility. The use of preformatted charts may expedite and facilitate the documentation process, thereby allowing more time for patient education. Additional studies are required to demonstrate this. Programmed charts also may facilitate research that uses chart reviews because these charts require uniform responses and improve completeness of documentation.
CONCLUSION
Programmed charts resulted in more complete documentation of ED records for a selected group of patients. The greatest improvements were in the recording of the history and physical examination. There was no detectable correlation between the use of programmed charts and patient health status outcomes, patient satisfaction with care, or patient perception of physician quality. There was a positive correlation with the use of preformatted charts and the patients' understanding of their discharge diagnosis, compliance with medications, and follow-up instructions. •
REFERENCES 1. Murphy JG, Jacobsen S: Assessing the quality of emergency care: The medical record versus patients' outcomes. Ann EmergMed 1984:13:158-165. 2. Switz D: The problem-oriented medical record. Arch/ntem Med 1976;136:1119-1123. 3. Sanazaro PJ, Worth RM: Concurrent quality assurance in hospital care. N EnglJ Med1974;298:1171-1177. 4. Fessel W J, Van Brunt EF: Assessing quality of care from the medical record. N Eng/J Med 1972;286:134-138. 5. Maran WT, Wiser TH, Nanda J, et al: Measuring medical residents chart documentation practices. J Med Educ 1988;63:859-865. 6. Traulhein J: Malpractice in the emergency department-- A review of 200 cases. Ann EmergMed 1984;13:709-711.
Address for reprints: Sheldon Jacobsen, MD, Emergency Services Department, Ground Floor Silverstein, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, Pennsylvania 19104.
i
L ii ! MAY 1992
21:5
ANNALS OF EMERGENCY MEDICINE
540177