A Needs Analysis for Computer-Based Telephone Triage in a Community AIDS Clinic S.B. Henry, R.N., D.N.Sc.l, J.G.Schreiner, M.A.1, D. Borchelt, R.N.2,and M.A. Musen, M.D., Ph.D.1
ISection on Medical Informatics Stanford University School of Medicine Stanford, CA 94305 2Immunocompromised Host Clinic Santa Clara Valley Medical Center San Jose, CA 95128 telephone-management activities into four categories based on review of 456 telephone-triage encounters: communication, coordination, education, and case management. Communication activities record information from outside sources; however, no decisions are made. Coordination activities include referrals, appointments, and arrangements for home care. Education activities require that the nurse interpret and explain information such as laboratory data or test results. Case-management activities involve organization and acquisition of patient information, exercise of judgment, and decision making, usually within the context of a clinical protocol. A survey questionnaire [2] identified a strong nurse preference for COSTAR (Computer Stored Ambulatory Record) versus the paper-based medical record, and also revealed several major benefits that were specifically relevant for telephone triage. The latter included handling telephone calls, recording information from telephone calls, and communicating with other staff members. We conducted a needs-analysis study to determine the appropriate computer-based strategies to support nursing telephone-triage management in communitybased AIDS clinics. For this study, a telephonetriage encounter is defined as the cluster of problemrelated activities initiated by a telephone call that identifies a patient problem or request. A single telephone call may initiate multiple data collection, assessment, intervention, communication, and documentation activities. Specifically, we addressed four research questions: (1) What is the current procedure for management of telephone-triage encounters in the clinic? (2) What is the nature of the telephone-triage task in a community-based AIDS clinic? (3)What are the data elements in telephone-triage encounters? and (4) What are the types of the data that are missing in the documentation for telephone-triage encounters?
ABSTRACT This study describes the complexity of the telephonetriage task in a community-based AIDS clinic. We identify deficiencies related to the data management for and documentation of the telephone-triage encounter, including inaccessibility of the medical record andfailure to document required data elements. Our needs analysis suggests five design criteria for a computer-based system that assists nurses with the telephone-triage task: (1) online accessibility of the medical record, (2) ability to move among modules of the medical record and the triage-encounter module, (3) ease of data entry, (4) compliance with standards for documentation, and (5) notification of the primary-care physician in an appropriate and timely
manner. TELEPHONE-TRIAGE DECISION MAKING
Several authors have identified features of telephone-triage decision making that distinguish it from diagnostic decision making [1-3]. The goal of telephone triage is to refer the patient to the appropriate health-care resources within an appropriate period of time. Patient safety is balanced against the efficient use of limited resources. In contrast to the complex strategies that may be associated with the management of a patient with a specific diagnosis, the outcomes of triage decision making are limited to one of four general conclusions: (1) the patient has no problem that requires management; (2) the patient has a problem that can be managed at home; (3) the patient has a problem that can be managed on an appointment basis; or (4) the patient has a problem that necessitates that he be seen immediately. Stoupa and Campbell [2] classified nursing
0195-4210/92/$5.00 0 1993 AMIA, Inc.
59
METHODS
duning 1991.
Our descriptive study design included both concurrent and retrospective data collection in one community-based AIDS clinic.
RESULTS
Our results relate to four specific areas: (1) the current procedure for management of telephone-triage encounters, (2) the nature of the telephone-triage task, (3) the data elements for telephone-triage documentation, and (4) the type and amount of missing data
Site The Immunocompromised Host (ICH) Clinic is the HIV/AIDS outpatient clinic of Santa Clara Valley Medical Center, a health-care facility operated by the county of Santa Clara, California. The clinic currently serves a population of approximately 350 patients annually; this number is increasing. The clinic staff comprises nine primary-care physicians, two registered nurses (RNs), one research nurse, one licensed vocational nurse (LVN), and one admissions clerk. The screening of telephone-triage calls is the responsibility of the two RNs in the clinic. Telephone triage as practiced in the ICH Clinic includes all the activities associated with coordination, education, and case management. Although communication activities occur, the information is recorded directly in the progress- note section of the clinic record; it is not documented as a telephone-triage encounter.
Current Procedure for Management of Telephone-Triage Encounters In the current procedure, each call is screened by an RN. Based on the nature of the call, she chooses one or more of the following activities: collect additional data from the medical record, advise the patient, refer patient to the emergency room, schedule a clinic appointment, and notify the physician. The written documentation of the encounter on the Triage Documentation form begins during the initial telephone call. The Triage Documentation form includes identification-related data, which are at the top of the current form, and problem-related data, which constitute the remainder of the form. However, documentation does not necessarily occur in that order. The RN often charts the reason for the call first; then, she collects and documents additional identification-related and problem-related data. To access pertinent clinical data related to the triage encounter, the RN requests the clinic chart from the Medical Records Department by completing a Medical Record Request form. She commonly obtains the patient's medical record number by entering the patient's name into the laboratory computer. The completed form is carried to Medical Records by the nurse or clerk, and the chart is retrieved by Medical Records personnel. This process takes approximately 10 minutes. If they are pertinent to the reason for the call, recent laboratory data are retrieved from the laboratory computer by the RN. The RN then documents additional data as appropriate to the reason for the call (e.g., current medications when the reported problem is nausea or vomiting, or recent hospitalization for Pneumocystis carinii pneumonia when the reported problem is shortness of breath). The nurse attaches the Triage Encounter form to the patient chart for the physician's perusal at the patient's next clinic visit or when the physician returns the patient's call. We could not estimate the actual time per triage encounter because, as our description illustrates, the encounter comprises various related activities, all of which are interrupted frequently by other patient-care activities in the clinic.
Design We conducted a descriptive study to examine the current procedure for telephone-triage encounters in the ICH Clinic, and to determine the data elements and sources of missing data related to triage encounters in the clinic. We used two data-collection methods: concurrent clinical observation and retrospective chart audit We observed the procedure for triage encounters on 5 nonconsecutive days during a 4-week period. We validated our observations of the procedure by discussions with the nurse manager in the ICH Clinic. We collected data from the patient charts related to patient demographics, HIV status, ICH clinic visits, and ICH telephone-triage encounters. We coded the triage encounters for source of call, problem or other reason for the call, RN action related to the call, and tpe of data missing in the documentation of the encounter.
Sample For the retrospective chart audit, we drew a random sample of 54 patients from the 275 patients seen in the ICH Clinic during the final quarter of 1991. The sample was predominantly male (93%) with a mean age of 34.6 years. The mean length of time as an ICH Clinic patient was 14.9 months, with a mean of 7.1 clinic visits and 1.8 triage encounters
60
Nature of the Telephone-Triage Task The characteristics of the clinic population provide the context for the triage task. Contextual variables of particular importance to the telephonetriage task in AIDS clinics are ethnicity and disease status. Forty-nine percent of the study sample were ethnic minorities: Latino (28 percent), Black (15 percent), Asian (3 percent) and other (3 percent). Fifty-four percent of the sample were diagnosed as having AIDS. The mean CD4 count was 211.8. There were 94 encounters for the 54 patients. The number of encounters per patient ranged from 0 to 13. Twenty-two patients (41 percent) had no documented triage encounters. The number of encounters was significantly correlated with the number of clinic visits (r = .52, p = .0001), but was not correlated with a marker of disease status, the CD4 count (r = -.14, p = .32). The most frequent initiator of the encounter was the patient (83 percent). The most frequent problem or other reason for the call was to report patient symptoms (82 percent). Other reasons for initiating a triage encounter included requests for medication refill (7 percent), for a telephone consultation with the primary care physician (7 percent), and for letters to meet legal or social-service needs (3 percent). A breakdown of the symptom-related data demonstrates the variety, complexity, and range of acuity of the symptoms reported by patients. A total of 110 symptoms were reported in the 77 symptomrelated triage encounters. The most frequently reported symptom was pain (14 percent), followed by fever (11 percent), and nausea or vomiting (10 percent). Some symptoms such as chest pain or shortness of breath required referral for immediate medical attention, whereas other symptoms such as constipation were managed by telephone advise or a follow-up appointment. Although some patients reported a single symptom, other patients reported multiple symptoms. RN actions related to the encounter included physician notification (63 percent), referral to the emergency room (20 percent), clinic appointment (12 percent), and advice only (5 percent).
data are entered as free text. The problem may be described in the caller's own words or may be summarized by the RN. Missing Data Data omitted from the encounter form included RN signature (23 percent) and RN actions (5 percent). One encounter form did not list the source and reason for the call; it listed only the RN action. DISCUSSION
Our needs analysis guided our selection of design goals for computer-based management of telephonetriage encounters in three community-based AIDS clinics in the San Francisco Bay area. We shall discuss each goal in turn, moving from the general goals to specific goals for correcting the deficiencies identified in the needs analysis. Online Accessibility of the Medical Record The inaccessibility of the medical record for the management of telephone triage has been documented in several investigations. In a Swedish study describing primary-care nurse's knowledge use and need during telephone consultations, Timpka and Arborelius categorized the dilemmas faced by the nurses by three types of knowledge needed: medicalscientific knowledge, social knowledge, and personal competence [4]. Of special relevance to the topic of our paper was the social dilemma of organizational breakdown, which was most commonly unavailability of the medical record. Stoupa and Campbell also identified inaccessibility of the medical record as a problem in the management of telephonetriage calls [2]. They documented an average delay of 45.2 minutes in receiving the requested medical record; 28 percent of the charts did not arrive during the day on which they were requested. Our data validate the need for online accessibility of the medical record. Our clinical observations documented the inefficiency in the current method of obtaining the medical record. In addition, we identified a broad array of symptom-related problems reported by the patients. Because the frequency of treatment-related symptoms in AIDS patients is high [5-7], immediate access to the current medication list and treatment plan is essential. We also described the ethnic diversity of the clinic population. An accessible medical record has the potential to serve as an adjunct for communicating with the subset of this population for whom English is not the primary language.
Data Elements The current Triage Documentation form includes identification-related and problem-related data elements. The identification-related data elements are date, time, patient name, patient telephone number, medical-record number, California Medicaid number, and physician. The problem-related data elements are problem or other reason for call, and RN actions taken. The identification-related data elements are entered into designated boxes. The problem-related
61
Maneuverability We described the nonlinear nature of the telephone-triage process. The RNs move among data collection, assessment, intervention, communication, and documentation activities during the telephonetriage encounter. Our data suggest that a computerbased approach that allows ease of navigation among modules of the medical record (e.g., problem list, progress notes and the Triage Encounter Documentation module) would be essential.
the 3-week evaluation study. Although there was an increase in the frequency of documentation of telephone calls, it fell short of the expected increase. Areas poorly documented were the reason for the call, the patient assessment, and the description of the intervention. The results of this study indicate that, although the computer-based approach was perceived as useful, a computer-based approach alone did not provide adequate support for the management of telephone-triage documentation. The authors identified the lack of a documentation standard as one reason for the poor documentation in the computerbased system. Our needs analysis identified deficiencies in the documentation of telephone-triage calls in the ICH Clinic. In contrast to Stoupa and Campbell's results, the reasons for the call were well documented. However, the interventions or actions were not documented in 5 percent of the encounters. RN signature was the most frequent type of missing datum (23 percent). Although problem or reason for the call, RN action, and RN signature are problemrelated data elements on the current Triage Encounter form, there is no established standard for documentation of telephone-triage encounters in the ICH clinic. Our needs analysis and the experience of Stoupa and Campbell suggest a dual approach: establishment of a documentation standard, and computer-based system documentation of the triage encounter including a reminder function in which the RN will be alerted if a required data element (e.g., problem, RN action, or signature) is not entered.
Ease of Data Entry We described the interactive nature of the telephone-triage task. The RN is simultaneously communicating with the caller and documenting the encounter. The need for ease of data entry must be balanced with the need to capture the data in a valid and useful forn. Shiffman described the design and implementation of a protocol-driven system for computer-assisted telephone triage in pediatrics, called Pediatric Telephone Protocols [3]. The program is based on information concerning a specific triage encounter and reference protocols to be followed for each call. The user records caller responses to items in the complaints stack with a mouse click. A major advantage of this system is the ease of data entry. Our needs analysis indicates that a single approach to data entry would be inadequate for management of the telephone-triage task in community-based AIDS clinics. We suggest a threepronged approach to data entry. We propose the transfer of identification-related data elements from the medical record so that the RN needs only to verify or update the data. We identified two problem-related data elements that were narrowly defined: source of call and RN actions. A mouse-based data-entry strategy may be the most appropriate for these data elements. We described the variety and complexity of the problems or reasons for the call. We also described the clustering of multiple problems within a single encounter. Free-text data entry from the keyboard would allow flexibility in several problemrelated areas. The data could be captured in the caller's own words or as summarized by the nurse. Multiple problems or reasons for the call could also be documented in one encounter.
Notification of the Physician in Appropriate and Timely Manner Our clinical observations indicated that the physicians consistently reviewed the telephone-triage notes prior to the patient's next clinic visit. The Triage Documentation forms were reviewed sooner if the nature of the problem required immediate physician attention. Our chart review, however, indicated that notification of the physician was documented only 63 percent of the time. We propose a computer-based notification system. The physician will be notified automatically of telephone-triage encounters for her patients by means of an alert message when she logs on. For problems requiring immediate physician attention, the current clinic procedure will continue. The physician will be notified by telephone or the patient will be referred to the emergency room.
Compliance with the Standard for Documentation Stoupa and Campbell described the use of a computer-based nursing module to document ambulatory care rendered by telephone within a COSTAR environment [2]. Fifty-six telephone calls were logged in the interactive nursing module during
CONCLUSIONS
Our needs analysis documented deficiencies in the
62
procedure for telephone-triage encounters, and validated our initial observations that a computerbased approach might be appropriate and effective. Specifically, it pointed out that the nurses faced difficulties in accessing relevant clinical data, and that a large number of data were missing in the documentation for telephone-triage encounters. In addition, our needs analysis guided us in the selection of design goals. Additional data to support the iterative collaborative design of the triage-encounter interface will be gathered in studies analyzing the conversational framework of telephone triage.
Acknowledgments We thank the members of the THERAPYHELPER team for their help and advice; we thank the administration and staff of the Immunocompromised Host Clinic for their cooperation with this project. This project was supported by grants HS-06330 from the Agency for Health Care Policy and Research and LM-07033 from the National Library of Medicine. Computing facilities were provided by the SUMEXAIM Resource, under grant NIH-LM-05208.
Reference [1] Corcoran, S , Narayan, S. and Moreland, H. "Thinking aloud" as a strategy to improve clinical decision making. Heart & Lung. 17:463-468, 1988.
[2] Stoupa, R. and Campbell, J.R. Documentation of ambulatory care rendered by telephone: Use of a computerized nursing module. In Miller, R.A. (ed), Proceedings of the Fourteenth Annual Symposium on Computer Applications in Medical Care, Los Alamitos, CA: IEEE Computer Society Press, 1990, 890-893. [31 Shiffman, R.N. Design and implementation of a system for computer-assisted telephone triage in pediatrics. In Miffer, R.A. (ed), Proceedings of the Fourteenth Annual Symposium on Computer Applications in Meegcal Care, Los Alamitos, CA: EEEE Computer Society Pfess, 1990, 826. [41 Timpka, T. and Arborelius, E. The primary-care nurse's dilemmas: A study of knowledge use and need during telephone consultations. Journal of Advanced Nursing. 15:1457-1465, 1990. [5] Holzemer, W.L. and Henry, S.B. Nursing care plans for people with HIV/AIDS: Confusion or consensus? Journal of Advanced Nursing. 16:257261, 1991. [61 Janson-Bierklie, S Holzemer, W.L., and Henry, S.B. Patient's perceptions of pulmonary problems and nursing interventions during hospitalization for Pneunwcystis carinii pneumonia. American Journal of Critical Carel (in press). [71 Henry, S.B., Holzemer, W.L., and Reilly, C.A. Nurses' perspectives on problems of hospitalized PCP patients: Implications for the development of a nursing taxonomy. Proceedings of the Fifteenth Annual Symposium on Computer Applications in Medical Care, Los Alamitos, CA: IEEE Computer Society Press, 1991, 177-181. ,
BRITISH MEDICAL JOURNAL. SA>TURDAY2 JUNE I $TH, I 878, THE TWO MEDICAL ACT AMENDMENT BILLS. t 858, Amendment Bill (fn . 2)" isnow before GiTns Medical Act, (Member foxr by Mr. Arthur MillsMills Whe House of Commons, endorsed boidney. Mr. Arthur
and Mr.
ISection on Medical Informatics Stanford University School of Medicine Stanford, CA 94305 2Immunocompromised Host Clinic Santa Clara Valley Medical Center San Jose, CA 95128 telephone-management activities into four categories based on review of 456 telephone-triage encounters: communication, coordination, education, and case management. Communication activities record information from outside sources; however, no decisions are made. Coordination activities include referrals, appointments, and arrangements for home care. Education activities require that the nurse interpret and explain information such as laboratory data or test results. Case-management activities involve organization and acquisition of patient information, exercise of judgment, and decision making, usually within the context of a clinical protocol. A survey questionnaire [2] identified a strong nurse preference for COSTAR (Computer Stored Ambulatory Record) versus the paper-based medical record, and also revealed several major benefits that were specifically relevant for telephone triage. The latter included handling telephone calls, recording information from telephone calls, and communicating with other staff members. We conducted a needs-analysis study to determine the appropriate computer-based strategies to support nursing telephone-triage management in communitybased AIDS clinics. For this study, a telephonetriage encounter is defined as the cluster of problemrelated activities initiated by a telephone call that identifies a patient problem or request. A single telephone call may initiate multiple data collection, assessment, intervention, communication, and documentation activities. Specifically, we addressed four research questions: (1) What is the current procedure for management of telephone-triage encounters in the clinic? (2) What is the nature of the telephone-triage task in a community-based AIDS clinic? (3)What are the data elements in telephone-triage encounters? and (4) What are the types of the data that are missing in the documentation for telephone-triage encounters?
ABSTRACT This study describes the complexity of the telephonetriage task in a community-based AIDS clinic. We identify deficiencies related to the data management for and documentation of the telephone-triage encounter, including inaccessibility of the medical record andfailure to document required data elements. Our needs analysis suggests five design criteria for a computer-based system that assists nurses with the telephone-triage task: (1) online accessibility of the medical record, (2) ability to move among modules of the medical record and the triage-encounter module, (3) ease of data entry, (4) compliance with standards for documentation, and (5) notification of the primary-care physician in an appropriate and timely
manner. TELEPHONE-TRIAGE DECISION MAKING
Several authors have identified features of telephone-triage decision making that distinguish it from diagnostic decision making [1-3]. The goal of telephone triage is to refer the patient to the appropriate health-care resources within an appropriate period of time. Patient safety is balanced against the efficient use of limited resources. In contrast to the complex strategies that may be associated with the management of a patient with a specific diagnosis, the outcomes of triage decision making are limited to one of four general conclusions: (1) the patient has no problem that requires management; (2) the patient has a problem that can be managed at home; (3) the patient has a problem that can be managed on an appointment basis; or (4) the patient has a problem that necessitates that he be seen immediately. Stoupa and Campbell [2] classified nursing
0195-4210/92/$5.00 0 1993 AMIA, Inc.
59
METHODS
duning 1991.
Our descriptive study design included both concurrent and retrospective data collection in one community-based AIDS clinic.
RESULTS
Our results relate to four specific areas: (1) the current procedure for management of telephone-triage encounters, (2) the nature of the telephone-triage task, (3) the data elements for telephone-triage documentation, and (4) the type and amount of missing data
Site The Immunocompromised Host (ICH) Clinic is the HIV/AIDS outpatient clinic of Santa Clara Valley Medical Center, a health-care facility operated by the county of Santa Clara, California. The clinic currently serves a population of approximately 350 patients annually; this number is increasing. The clinic staff comprises nine primary-care physicians, two registered nurses (RNs), one research nurse, one licensed vocational nurse (LVN), and one admissions clerk. The screening of telephone-triage calls is the responsibility of the two RNs in the clinic. Telephone triage as practiced in the ICH Clinic includes all the activities associated with coordination, education, and case management. Although communication activities occur, the information is recorded directly in the progress- note section of the clinic record; it is not documented as a telephone-triage encounter.
Current Procedure for Management of Telephone-Triage Encounters In the current procedure, each call is screened by an RN. Based on the nature of the call, she chooses one or more of the following activities: collect additional data from the medical record, advise the patient, refer patient to the emergency room, schedule a clinic appointment, and notify the physician. The written documentation of the encounter on the Triage Documentation form begins during the initial telephone call. The Triage Documentation form includes identification-related data, which are at the top of the current form, and problem-related data, which constitute the remainder of the form. However, documentation does not necessarily occur in that order. The RN often charts the reason for the call first; then, she collects and documents additional identification-related and problem-related data. To access pertinent clinical data related to the triage encounter, the RN requests the clinic chart from the Medical Records Department by completing a Medical Record Request form. She commonly obtains the patient's medical record number by entering the patient's name into the laboratory computer. The completed form is carried to Medical Records by the nurse or clerk, and the chart is retrieved by Medical Records personnel. This process takes approximately 10 minutes. If they are pertinent to the reason for the call, recent laboratory data are retrieved from the laboratory computer by the RN. The RN then documents additional data as appropriate to the reason for the call (e.g., current medications when the reported problem is nausea or vomiting, or recent hospitalization for Pneumocystis carinii pneumonia when the reported problem is shortness of breath). The nurse attaches the Triage Encounter form to the patient chart for the physician's perusal at the patient's next clinic visit or when the physician returns the patient's call. We could not estimate the actual time per triage encounter because, as our description illustrates, the encounter comprises various related activities, all of which are interrupted frequently by other patient-care activities in the clinic.
Design We conducted a descriptive study to examine the current procedure for telephone-triage encounters in the ICH Clinic, and to determine the data elements and sources of missing data related to triage encounters in the clinic. We used two data-collection methods: concurrent clinical observation and retrospective chart audit We observed the procedure for triage encounters on 5 nonconsecutive days during a 4-week period. We validated our observations of the procedure by discussions with the nurse manager in the ICH Clinic. We collected data from the patient charts related to patient demographics, HIV status, ICH clinic visits, and ICH telephone-triage encounters. We coded the triage encounters for source of call, problem or other reason for the call, RN action related to the call, and tpe of data missing in the documentation of the encounter.
Sample For the retrospective chart audit, we drew a random sample of 54 patients from the 275 patients seen in the ICH Clinic during the final quarter of 1991. The sample was predominantly male (93%) with a mean age of 34.6 years. The mean length of time as an ICH Clinic patient was 14.9 months, with a mean of 7.1 clinic visits and 1.8 triage encounters
60
Nature of the Telephone-Triage Task The characteristics of the clinic population provide the context for the triage task. Contextual variables of particular importance to the telephonetriage task in AIDS clinics are ethnicity and disease status. Forty-nine percent of the study sample were ethnic minorities: Latino (28 percent), Black (15 percent), Asian (3 percent) and other (3 percent). Fifty-four percent of the sample were diagnosed as having AIDS. The mean CD4 count was 211.8. There were 94 encounters for the 54 patients. The number of encounters per patient ranged from 0 to 13. Twenty-two patients (41 percent) had no documented triage encounters. The number of encounters was significantly correlated with the number of clinic visits (r = .52, p = .0001), but was not correlated with a marker of disease status, the CD4 count (r = -.14, p = .32). The most frequent initiator of the encounter was the patient (83 percent). The most frequent problem or other reason for the call was to report patient symptoms (82 percent). Other reasons for initiating a triage encounter included requests for medication refill (7 percent), for a telephone consultation with the primary care physician (7 percent), and for letters to meet legal or social-service needs (3 percent). A breakdown of the symptom-related data demonstrates the variety, complexity, and range of acuity of the symptoms reported by patients. A total of 110 symptoms were reported in the 77 symptomrelated triage encounters. The most frequently reported symptom was pain (14 percent), followed by fever (11 percent), and nausea or vomiting (10 percent). Some symptoms such as chest pain or shortness of breath required referral for immediate medical attention, whereas other symptoms such as constipation were managed by telephone advise or a follow-up appointment. Although some patients reported a single symptom, other patients reported multiple symptoms. RN actions related to the encounter included physician notification (63 percent), referral to the emergency room (20 percent), clinic appointment (12 percent), and advice only (5 percent).
data are entered as free text. The problem may be described in the caller's own words or may be summarized by the RN. Missing Data Data omitted from the encounter form included RN signature (23 percent) and RN actions (5 percent). One encounter form did not list the source and reason for the call; it listed only the RN action. DISCUSSION
Our needs analysis guided our selection of design goals for computer-based management of telephonetriage encounters in three community-based AIDS clinics in the San Francisco Bay area. We shall discuss each goal in turn, moving from the general goals to specific goals for correcting the deficiencies identified in the needs analysis. Online Accessibility of the Medical Record The inaccessibility of the medical record for the management of telephone triage has been documented in several investigations. In a Swedish study describing primary-care nurse's knowledge use and need during telephone consultations, Timpka and Arborelius categorized the dilemmas faced by the nurses by three types of knowledge needed: medicalscientific knowledge, social knowledge, and personal competence [4]. Of special relevance to the topic of our paper was the social dilemma of organizational breakdown, which was most commonly unavailability of the medical record. Stoupa and Campbell also identified inaccessibility of the medical record as a problem in the management of telephonetriage calls [2]. They documented an average delay of 45.2 minutes in receiving the requested medical record; 28 percent of the charts did not arrive during the day on which they were requested. Our data validate the need for online accessibility of the medical record. Our clinical observations documented the inefficiency in the current method of obtaining the medical record. In addition, we identified a broad array of symptom-related problems reported by the patients. Because the frequency of treatment-related symptoms in AIDS patients is high [5-7], immediate access to the current medication list and treatment plan is essential. We also described the ethnic diversity of the clinic population. An accessible medical record has the potential to serve as an adjunct for communicating with the subset of this population for whom English is not the primary language.
Data Elements The current Triage Documentation form includes identification-related and problem-related data elements. The identification-related data elements are date, time, patient name, patient telephone number, medical-record number, California Medicaid number, and physician. The problem-related data elements are problem or other reason for call, and RN actions taken. The identification-related data elements are entered into designated boxes. The problem-related
61
Maneuverability We described the nonlinear nature of the telephone-triage process. The RNs move among data collection, assessment, intervention, communication, and documentation activities during the telephonetriage encounter. Our data suggest that a computerbased approach that allows ease of navigation among modules of the medical record (e.g., problem list, progress notes and the Triage Encounter Documentation module) would be essential.
the 3-week evaluation study. Although there was an increase in the frequency of documentation of telephone calls, it fell short of the expected increase. Areas poorly documented were the reason for the call, the patient assessment, and the description of the intervention. The results of this study indicate that, although the computer-based approach was perceived as useful, a computer-based approach alone did not provide adequate support for the management of telephone-triage documentation. The authors identified the lack of a documentation standard as one reason for the poor documentation in the computerbased system. Our needs analysis identified deficiencies in the documentation of telephone-triage calls in the ICH Clinic. In contrast to Stoupa and Campbell's results, the reasons for the call were well documented. However, the interventions or actions were not documented in 5 percent of the encounters. RN signature was the most frequent type of missing datum (23 percent). Although problem or reason for the call, RN action, and RN signature are problemrelated data elements on the current Triage Encounter form, there is no established standard for documentation of telephone-triage encounters in the ICH clinic. Our needs analysis and the experience of Stoupa and Campbell suggest a dual approach: establishment of a documentation standard, and computer-based system documentation of the triage encounter including a reminder function in which the RN will be alerted if a required data element (e.g., problem, RN action, or signature) is not entered.
Ease of Data Entry We described the interactive nature of the telephone-triage task. The RN is simultaneously communicating with the caller and documenting the encounter. The need for ease of data entry must be balanced with the need to capture the data in a valid and useful forn. Shiffman described the design and implementation of a protocol-driven system for computer-assisted telephone triage in pediatrics, called Pediatric Telephone Protocols [3]. The program is based on information concerning a specific triage encounter and reference protocols to be followed for each call. The user records caller responses to items in the complaints stack with a mouse click. A major advantage of this system is the ease of data entry. Our needs analysis indicates that a single approach to data entry would be inadequate for management of the telephone-triage task in community-based AIDS clinics. We suggest a threepronged approach to data entry. We propose the transfer of identification-related data elements from the medical record so that the RN needs only to verify or update the data. We identified two problem-related data elements that were narrowly defined: source of call and RN actions. A mouse-based data-entry strategy may be the most appropriate for these data elements. We described the variety and complexity of the problems or reasons for the call. We also described the clustering of multiple problems within a single encounter. Free-text data entry from the keyboard would allow flexibility in several problemrelated areas. The data could be captured in the caller's own words or as summarized by the nurse. Multiple problems or reasons for the call could also be documented in one encounter.
Notification of the Physician in Appropriate and Timely Manner Our clinical observations indicated that the physicians consistently reviewed the telephone-triage notes prior to the patient's next clinic visit. The Triage Documentation forms were reviewed sooner if the nature of the problem required immediate physician attention. Our chart review, however, indicated that notification of the physician was documented only 63 percent of the time. We propose a computer-based notification system. The physician will be notified automatically of telephone-triage encounters for her patients by means of an alert message when she logs on. For problems requiring immediate physician attention, the current clinic procedure will continue. The physician will be notified by telephone or the patient will be referred to the emergency room.
Compliance with the Standard for Documentation Stoupa and Campbell described the use of a computer-based nursing module to document ambulatory care rendered by telephone within a COSTAR environment [2]. Fifty-six telephone calls were logged in the interactive nursing module during
CONCLUSIONS
Our needs analysis documented deficiencies in the
62
procedure for telephone-triage encounters, and validated our initial observations that a computerbased approach might be appropriate and effective. Specifically, it pointed out that the nurses faced difficulties in accessing relevant clinical data, and that a large number of data were missing in the documentation for telephone-triage encounters. In addition, our needs analysis guided us in the selection of design goals. Additional data to support the iterative collaborative design of the triage-encounter interface will be gathered in studies analyzing the conversational framework of telephone triage.
Acknowledgments We thank the members of the THERAPYHELPER team for their help and advice; we thank the administration and staff of the Immunocompromised Host Clinic for their cooperation with this project. This project was supported by grants HS-06330 from the Agency for Health Care Policy and Research and LM-07033 from the National Library of Medicine. Computing facilities were provided by the SUMEXAIM Resource, under grant NIH-LM-05208.
Reference [1] Corcoran, S , Narayan, S. and Moreland, H. "Thinking aloud" as a strategy to improve clinical decision making. Heart & Lung. 17:463-468, 1988.
[2] Stoupa, R. and Campbell, J.R. Documentation of ambulatory care rendered by telephone: Use of a computerized nursing module. In Miller, R.A. (ed), Proceedings of the Fourteenth Annual Symposium on Computer Applications in Medical Care, Los Alamitos, CA: IEEE Computer Society Press, 1990, 890-893. [31 Shiffman, R.N. Design and implementation of a system for computer-assisted telephone triage in pediatrics. In Miffer, R.A. (ed), Proceedings of the Fourteenth Annual Symposium on Computer Applications in Meegcal Care, Los Alamitos, CA: EEEE Computer Society Pfess, 1990, 826. [41 Timpka, T. and Arborelius, E. The primary-care nurse's dilemmas: A study of knowledge use and need during telephone consultations. Journal of Advanced Nursing. 15:1457-1465, 1990. [5] Holzemer, W.L. and Henry, S.B. Nursing care plans for people with HIV/AIDS: Confusion or consensus? Journal of Advanced Nursing. 16:257261, 1991. [61 Janson-Bierklie, S Holzemer, W.L., and Henry, S.B. Patient's perceptions of pulmonary problems and nursing interventions during hospitalization for Pneunwcystis carinii pneumonia. American Journal of Critical Carel (in press). [71 Henry, S.B., Holzemer, W.L., and Reilly, C.A. Nurses' perspectives on problems of hospitalized PCP patients: Implications for the development of a nursing taxonomy. Proceedings of the Fifteenth Annual Symposium on Computer Applications in Medical Care, Los Alamitos, CA: IEEE Computer Society Press, 1991, 177-181. ,
BRITISH MEDICAL JOURNAL. SA>TURDAY2 JUNE I $TH, I 878, THE TWO MEDICAL ACT AMENDMENT BILLS. t 858, Amendment Bill (fn . 2)" isnow before GiTns Medical Act, (Member foxr by Mr. Arthur MillsMills Whe House of Commons, endorsed boidney. Mr. Arthur
and Mr.
