Evaluating Patient Compliance With Nurse Advice Line Recommendations and the Impact on Healthcare Costs

, , , ,
The American Journal of Managed Care, August 2004, Volume 10, Issue 8

Objective: To explore the effect of telephone triage and advicelines in uninsured and managed care populations served by a safetynet system and to document the relationship between thepatient's initial plan for healthcare, the nurse recommendation,and the patient's subsequent healthcare action.

Study Design: Prospective telephone survey.

Patients and Methods: Of 1538 calls to a nurse advice line in a28-day period, 710 (46%) callers were selected to be surveyed. Ofthose, 278 (39%) were surveyed by telephone within 7 days oftheir call to assess patient compliance with recommendations, thepatient's actual healthcare actions, and their satisfaction with theservice.

Results: Patients' reported actions were classified as either (1)home care (46%), (2) clinic visit (27%), or (3) hospital visit (27%).Seventy percent of patients complied with nurse advice line recommendations.Most patient actions (68%) differed from their originalhealthcare plan, with many (46%) choosing a lower intensityof care. Changes from patients' original healthcare plans had apotential annual net savings of $322 249.

Conclusion: The simple act of calling a nurse triage and adviceline corresponds with a change in the reported actions of uninsuredand managed care patients and a potential reduction in coststo the safety net system providing their healthcare.

(Am J Manag Care. 2004;10:534-542)

Healthcare systems throughout the United Stateshave created telephone triage and advice linesto recommend to patients the most appropriateintensity of care.1 It is estimated that more than 100million people in the United States have access to telephonetriage and advice lines through a particular hospitalor health maintenance organization (HMO).2 Theutility of these services to both the healthcare systemand the patient assumes compliance with the adviceoffered and that the advice offered is cost effective.However, few studies have assessed whether the adviceprovided is actually followed by patients or whether itactually helps to decrease healthcare costs.

Appropriate utilization of facilities allows healthcaresystems to reduce expenditures for uncompensatedcare and to better serve managed care patients withinfinancial limitations. Telephone triage, decision support,and advice lines may assist in cost containment ifpatients follow the recommendations given.

Studies have reported patient compliance rangingfrom 4.5% to 88.2% depending upon the intensity of carerecommended and the criteria by which compliance wasdetermined.3-8 Most studies described triage and adviceservices for pediatric patient populations and parentalcompliance with recommendations provided by nursesor pediatricians.3,5,7,8 The only 2 studies to examine theutility of nurse advice line services for general patientpopulations consisted largely of insured patients.4,6

Such an assessment is important because patientsusing safety net systems are often uninsured or coveredby Medicaid. A successful nurse triage and advice lineservice could reduce costs and improve access to carefor these populations. The following study describes theeffect of a nurse triage and advice line service on patienthealthcare actions in a safety net system by investigatingthe caller's original plan for healthcare, the nurserecommendation, and the patient's self-reported subsequentaction.


The Denver Health NurseLine was established onJune 2, 1997, to provide triage care recommendationsto patients of Denver Health, an urban safety net healthnetwork serving a population of approximately 500 000people. At the time of this study, the NurseLine receiveda total of 30 000 patient contacts per year. Patients calla local telephone number to contact the NurseLine, andmessages in Spanish or English direct those with emergencyconcerns to call 911. Callers then speak witha nurse. Once connected, the nurse gathers patienthistory and demographic data before triage. A languageline service is used to assist non—English-speakingcallers. Although the same demographic data is capturedand the same triage is provided, non—English-speakingcallers were not surveyed for this study due tothe language barrier. With the assistance of a commercialsoftware program, Centramax (McKesson HBOC,San Francisco, Calif), the nurse makes an assessmentand a recommendation for appropriate care. Before thecall ends, patients are asked what they would have donehad they not utilized the NurseLine.

Methodology for this study was based on a preliminarysurvey as well as other reported studies.3,4,9 This studywas approved by the Colorado Multiple InstitutionalReview Board of the University of Colorado HealthSciences Center (Denver). A random-number generatorselected callers from reports of all triage calls to DenverHealth NurseLine on each of the 28 days betweenNovember 26, 2000, and December 23, 2000, as stratifiedby shift: day (7:00 AM—2:59 PM), evening(3:00 PM—10:59 PM), and night (11:00 PM—6:59 AM). Theratio of calls to be sampled per shift was determined usinga preliminary tally of all triage calls received duringOctober 2000: 678 calls during the day shift, 992 callsduring the evening shift, and 188 calls during the nightshift (a 3:5:1 ratio). Based on this information, approximately25 calls were randomly selected per 24-hour day:8 (32%) day, 14 (56%) evening, and 3 (12%) night shift.At least 3 attempts were made to contact callers over 3different days with a minimum of 2 different shifts.

Within 7 days of their NurseLine calls, randomlyselected callers were contacted by a trained researchassistant who initiated telephone surveys using a standardized5-item questionnaire (Appendix). The questionnairewas designed to assess the following:(1) caller's subsequent action for care, (2) satisfactionwith the NurseLine service, (3) satisfaction with thenurse's recommendations, (4) perception of wait timefor service, and (5) intention to utilize the service in thefuture. The relationship between caller's original plan,the NurseLine recommendations, and caller's subsequentaction was examined. Plans, recommendations,and actions were categorized into 3 levels of care: homecare, clinic visit, and hospital visit.

Home care


Hospital visit

was considered the lowest intensity ofcare and included all actions not involving a visit to ahealthcare agency (bed rest, stay home, etc). was the next highest intensity of care and includedall nonemergent actions for healthcare (see doctor,call doctor, go to clinic, etc). , the highestintensity of care, included all emergent actions forhealthcare (go to emergency department, call 911, go tohospital, etc). For example, if a caller's original plan wasto go to their community clinic and the nurse recommendedbed rest with fluids, then the original plan wascategorized as clinic visit, and the recommendation wascategorized as home care. The care recommended wasof lower intensity than what the patient had originallyplanned. If the subsequent patient survey indicated thathome care was followed, a lower intensity of care had infact been chosen, and there was concordance of nurserecommendation and patient action. However, therewas discordance between the caller's initial plan and thenurse's recommendation. This survey reveals the patternof behavior from the caller's original plan to nurserecommendation to subsequent action.

The system savings potential of the nurse triage andadvice line service was calculated by extending the dataregarding alterations in healthcare decisions from thesample to all users. Average charges for home care($0.00), a clinic visit ($137.08), and a hospital emergencydepartment visit ($969.01) in our system wereused to estimate cost savings associated with changesfrom the patient's original healthcare plan to their subsequentaction as a result of nurse advice line recommendations.

In an attempt to validate the self-reported actions ofcallers, a patient search was conducted within theDenver Health clinical database. The search criteriaincluded a patient visit for either inpatient or outpatientcare within 7 days of a call to the NurseLine. This confirmedthe statements of those who said they soughtcare at a clinic or hospital and those who stated thatthey did not seek care outside of the home. This searchcould neither confirm nor exclude any patient visitsoutside of the Denver Health network.

Survey respondents were asked to rate their satisfactionwith the NurseLine service using a scale rangingfrom 1 (extremely dissatisfied) to 10 (extremely satisfied).Respondents were also asked to rate their satisfactionwith the nurse's recommendations using thesame 10-point scale. To gauge the caller's perception ofNurseLine staff accessibility, respondents were asked ifthey waited long to speak to a nurse (yes or no) and toestimate how many minutes they had waited. Respondentswere also asked if they would use the NurseLineservice again.

Overall patient compliance was analyzed in 3 categories:(1) the caller's original plan versus the nurse'srecommendation, (2) the nurse's recommendation versusthe caller's self-reported subsequent action, and (3)the caller's original plan versus the caller's self-reportedsubsequent action. These patterns illustrate the directionof any deviance from the original plan to the recommendationto the subsequent action based upon thelevel of intensity of care.


Because these are categorical data, the &#967;2 test forindependence is used to discern any differences indemographics between people who completed the surveyand people who did not (statistically significant if < .05). Additional descriptive statistics provide asummary analysis of caller satisfaction,wait time, and intention to utilize the NurseLine again.


During the 28-day study period, the NurseLine received 1538 triage calls,of which 710 cases (46%) were randomly selected. Of those 710, 278individuals (39%) completed the survey. The remaining 432 individuals(61%) did not participate for a variety of reasons (Figure 1).



The majority (67%) of patients were female and either did not have insurance(33.8%) or relied upon Medicaid programs (46%) for medical payment(Table 1). Managed care patients comprised 24% of those surveyed. Patientage ranged from 4 days to 85 years, with an average patient age of 23 years(SD &#177; 20). Although detailed information for the nonresponders is not presented,there was no significant difference in shift ratio, age, or genderbetween those who participated in the survey and those who did not( > 0.05). Nor was there a significant difference in shift ratio, age, or genderbetween those in managed care and those in other insurance statusgroups ( > 0.05).

In order to make comparisonsbased on the relationship between thecaller's original plan, nurse's recommendation,and patient's subsequentaction, only the 266 respondents withcomplete data for all three outcomeswere analyzed (96%).

The majority of callers (56%)reported that they would have gone toa clinic for care if they had not usedthe NurseLine service (Figure 2).Another 29% reported that they wouldhave gone to the hospital, and 16%stated that they would have practicedself-care at home (Figure 2). The estimatedhealthcare charges for the original plan of callerswas $93 933 (Table 2).

The NurseLine recommended a clinic visit as theappropriate intensity of care in 44% of cases, referral toa hospital in 20% of cases and home care in 36% ofcases (Figure 2). There was concordance of the caller'soriginal plan and the NurseLine recommendation in101 cases (38%), with the greatest agreement for clinicvisit. More importantly, the NurseLine recommendationwas for a lower intensity of care in 116 cases (44%)and for a higher intensity of care in 49 cases (18%;Figure 2). The estimated charges for healthcareactions recommended by the NurseLine service were$67 533, represents $26 400 in charges avoided incomparison to the caller's original plan (Table 2).Charges for the lower-intensity care recommended byNurseLine are less than those for higher-intensity careby a 2:1 ratio.

Of the self-reported caller actions, 187 (70%) were inconcordance with the documented nurse recommendation(Figure 2). Compliance with recommendationswas highest for home care (86/95 = 91%), followed byhospital visit (40/53 = 75%) and clinic visit (61/118 =52%). Thus a majority of nurse recommendations wereheeded by callers. However, when asked "Did you followthe nurse's recommendation?" 94% of the callerssaid "Yes." This difference may indicate that the callermisunderstood or could not remember the appropriatelevel of care recommended; it may also indicate poordocumentation or communication by the nurse. Forthe discordant cases, it is noteworthy that 43 callers(16%) chose a lower intensity of care and 36 (14%)chose a higher intensity of care than recommended.Estimated charges for the NurseLine recommendationswere $11 831 less than the $79 365 in chargesestimated for subsequent patient actions. The totalcharges for subsequent patient actions are $14 568less than those associated with the caller's original plan(Table 2).

Where possible, we corroborated the self-reportedactions (home care, clinic visit, or hospital visit) of the278 survey respondents with our in-patient and outpatientvisit databases. Only if a patient received documentedcare in a clinic or the hospital within 7 days ofcalling did we consider those self-reported subsequentactions to have been supported. We also reasoned thatthe absence of documented use of services indicatedhome care since our institution is the most likelyprovider of healthcare for this patient population. Ourrecords indicated that 93 patients visited a clinic or hospitalvisit in our system within 7 days of calling theNurseLine. We used ICD-9 codes to determine that 83of these 93 patients correctly self-reported a clinic orhospital visit related to their NurseLine call. The other10 patients reported their subsequent actions incorrectly(patient went to a clinic though they reported ahospital visit or home care, or the patient visited thehospital though they reported a clinic visit or homecare). With home care represented by no documenteduse of services within 7 days ofa NurseLine call, 116 of theremaining 176 callers correctlyself-reported home carerelated to their call. The other60 patients reported their subsequentactions incorrectly (asvisiting either a clinic or a hospital).This indicated that 74%of callers (199 of 278) reportedtheir subsequent healthcareactions consistent with inpatientand outpatient databases(patient records).

The relationships betweencallers' original plans forhealthcare and their self-reportedsubsequent actionsare summarized in Table 3. In31% of the cases, the callerreported pursuing the sameintensity of care as originallyplanned. However, over twothirds of callers (69%) altered their plan of action aftercalling the NurseLine. Most (47%) indicated that theirsubsequent action was of lower intensity than originallyplanned, while the remainder (22%) reported pursuingcare of higher intensity than what they initiallyplanned.

Of particular note are those patients who most radicallychanged their original plan for healthcare afteraccessing the NurseLine service (Table 3). For example,73 out of 148 patients (49%) who were planning on a clinicvisit decided on home care, and 53 out of 76 patients(70%) who planned to go to the hospital instead went to aclinic or utilized home care. Conversely, 14 out of 42patients (33%) who intended home care actually soughthospital care, and 35 out of 148 patients (24%) whointended to utilize clinic services sought hospital care.

The majority of respondents (89%) rated their overallsatisfaction with the NurseLine service at 7 or higher(Figure 3). Similarly, 91% rated their overall satisfactionwith the nurse recommendation at 7 or higher. Only 44(16%) indicated they had to "wait long." to speak with anurse, and the estimated wait time ranged from 0 to 80minutes with a mean of 5 minutes (SD &#177; 9.6). Mostrespondents (98%) said they would use the service again.



Further comparison of managed care and nonmanagedcare patients (Table 4) showed that the 2 groupsdiffered little in intensity of care originally planned bythe caller, the caller's subsequent action, and the percentageof those who reported compliance with thenurse recommendation ( > 0.05). The NurseLine servicerecommended different intensities of care for the 2groups: a higher percentage of managed care patientsreceived a clinic visit recommendation, and home carewas recommended to a higher percentage of nonmanagedcare patients ( < 0.05). There was no differencein the percentages of the 2 groups provided a hospitalvisit recommendation.


An important goal of the healthcare system fromboth a patient safety and cost efficiency perspective isto assure that patients receive the right care, in theright place, at the right time. One effort to achieve thisgoal has been the development of nurse triage andadvice line services. The utility of these servicesdepends upon patients' willingness to use such a service,the recommendations given, and patient compliancewith recommendations.

Despite the importance of this information, few studieshave examined patient compliance with nurse triageand advice service recommendations.3-8 Most of thesestudies examined triage services for pediatric patients orfor populations with some form of health insurance. Bothof the published studies examining the effect of nursetelephone triage in a general patient population servedpopulations with fewer than 16% uninsured patients.4,6Our survey population consisted mostly of the uninsuredor those utilizing government medical programs such asMedicare and Medicaid in a safety net healthcare system.For comparison, however, our system also includedpatients in managed care plans.

It is necessary to examine the utility ofnurse triage and advice services for the underinsuredor uninsured population for severalreasons. First, these patients often encountereconomic, transportation, and cultural obstaclesthat impede healthcare access.10 Areadily available nurse triage service couldreduce these barriers.Second, these patients often lack a primarycare physician and utilize the emergencydepartment for care.Most visits may notrequire this high-costcare venue for treatment.Third, althoughuninsured patients oftendefer needed care,11they may be less likelyto do so if advised by anurse to pursue care. Fourth, thesafety net systems that provide careto these populations are often financiallyfragile and, to remain viable,must maximize the efficiency of providingcare.12 Finally, Medicaid budgetsin a majority of states are underduress, and improved efficiency ofcare could benefit state budgets andthose who provide services by controllingcosts through appropriate utilizationof facilities and resources.

We studied the utilization and outcomesof a nurse triage and adviceservice at Denver Health, Colorado'sprincipal safety net healthcare institution.Within 3 years of its establishment,the NurseLine service wasreceiving 30 000 calls per year,demonstrating that both managedcare and uninsured patients will utilizesuch services if they are available.With 90% or more of service users reportingsatisfaction with the service, the nurse recommendationand would use the service again, this is an effective way to help all patients make appropriate healthcaredecisions.

The recommendations provided by nurses or physiciansproviding telephone triage advice were very similaracross studies despite differences in patientinsurance status. We found, as did the other studies,that nonurgent care at a clinic/physician's office or selfcareat home were most often recommended, with lessthan 20% of patients directed to seek urgent care at ahospital.5-8 As in other studies, 90% or more of our surveyrespondents reported following triage recommendations.3,4,7 However, our examination of inpatient andoutpatient records determined that the accuracy of self-reportedpatient actions for healthcare was approximately70%, which demonstrates the challenge ofrelying solely upon recollections of those surveyed.

Patient compliance with our triage service variedaccording to the intensity of care recommended, withthe least compliance associated with recommendationsfor nonurgent care at a clinic or physician office.Patients were much more likely to comply with recommendationsto go to the hospital or for home care.Other studies have reported similar findings for patientcompliance 3,5,6,8 However, no other study has describedwhat healthcare actions patients would have takenwithout the triage and advice service. We demonstratedthat, in addition to overall patient compliance of 70%,our nurse triage service was able to change the behaviorof 69% of patients. After speaking to a nurse, almost halfof these patients decreased the intensity of care theypursued, while approximately a fourth increased theintensity. This finding may indicate that patients arebeing directed to a more appropriate intensity of carefor their health concern. It also suggests that somepatients are being directed to pursue urgent care whomight have remained home or waited for a clinicappointment. Although the study was not designed todetermine if a higher intensity of care was really needed,a nurse's recommendation suggests it may well havebeen. Preventing the delay in care for these patientsmay ultimately prevent more costly care at a later date.

We were surprised to find a difference in nurse recommendationsto different patient populations: a clinicvisit was more commonly recommended to managedcare patients than to nonmanaged care patients. It isunlikely that this relates to the poverty of patients sincethe majority of managed care patients in this study areenrolled in a Medicaid program. Although nurses knowthe insurance status of patients when interviewed theydid not believe it altered their recommendations. Oneexplanation could be the different utilization of the serviceby the 2 populations. A managed care population mayknow more about healthcare issues and have additionalresources (books, Internet, etc) to assist with their healthconcerns. Nonmanaged care patients may have fewerresources and may need to contact the service with anyhealth concern. The difference or differences betweenpatient populations requires further investigation.

Our study describes the potential savings from moreeffective utilization of facilities by patients who use atriage and advice line. One limitation is the use ofpatient charges instead of actual-care costs to determinethe degree of this impact on healthcare expenditures.However, any decrease in utilization of resourcesin our maximally utilized not-for-profit, single-budgetsystem allows us to provide care, including uncompensatedcare, to others. Therefore, with our patient mixand fixed cost-recovery constraints, any charges avoidedfor uncompensated and managed care patientsreduce expenditures and thereby increase capacity inthe system, so that more healthcare can be delivered.

Our study indicated that the actions 266 patientstook after calling the NurseLine service ultimately saveda total of $14 568 in charges that would have beenincurred had they followed their original healthcareplan. This represents an average of $54.77 in chargesavoided per patient. Multiplying this average by theNurseLine's 30 000 annual patient contacts yields$1 643 100 in charges avoided. Multiplication by ourconsolidated ratio of cost to charges (59%) equates toa savings of $969 429. The NurseLine service has anoperating budget of $647 000 per year but no associatedpatient charges. Subtracting the service budgetleaves an estimated net cost savings of $322 429annually for the Denver Health system.

There are clear shortcomings in equating the potentialpatient-care savings from utilization of a nursetriage and advice service to actual savings. First, extrapolatingfrom the survey group to the entire population ofusers may not be valid. Second, if no actual personnelor supplies were decreased, these potential savings tothe system may not be realized unless perhaps moreuncompensated care can be provided with thoseresources. This problem is inherent in any estimate ofcost avoidance. Even with these limitations, the systemis likely to realize some meaningful savings.

While this study supports the effectiveness of phonetriage in getting a patient to the right venue of care, itdoes have limitations. Measuring patients' actual complianceand obtaining complete financial data for whatcosts are incurred in their treatment are problematic.Sound compliance measurement requires accuratepatient recollections of nurse recommendations, well-definedlevels of healthcare, ability to survey all serviceusers, and ability to validate all patient self-reportedactions. The latter also impacts the determination ofactual-dollar savings. Nonetheless, this study suggeststhat nurse triage and advice services, used by populationsserved by a safety net institution, provide a newlow-cost venue for access to care. Such services havethe potential to improve appropriateness of venues ofhealthcare for such populations and thereby improvethe financial health of the institutions that provide carefor these patients.

From Rocky Mountain Poison and Drug Center, Denver, Colo (GMB, JLG, PB, RCD)and Denver Health NurseLine, Denver Health and Hospital Authority, Denver, Colo (DS,PG, RCD).

This project was funded internally by Denver Health.

Address correspondence to: Gregory M. Bogdan, PhD, Rocky Mountain Poison andDrug Center&#150;Denver Health, 777 Bannock Street, Mail Code 0180, Denver, CO 80204.Email: greg.bogdan@rmpdc.org

Coll Rev

1. Wright SL. The primary care practice in transition: strategies for managing theevolution from fee for service to prepaid care. . 1998;15:5-26.


2. Nurse telephone-triage. . 2001;357:323.


3. Kempe A, Luberti AA, Hertz AR, et al. Delivery of pediatric after-hours care bycall centers: a multicenter study of parental perceptions and compliance.. 2001;108(6):1359-1360.

Am J ManagedCare

4. Moore JD, Saywell RM, Thakker N, Jones TA. An analysis of patient compliancewith nurse recommendations from an after-hours call center. . 2002;8(4):343-351.


5. Lee TJ, Guzy J, Johnson D, Woo H, Baraff LJ. Caller satisfaction with after-hourtelephone advice: nurse advice service versus on-call pediatricians. .2002;110(5):865-872.

Med DecisMaking

6. O'Connell JM, Towles W, Yin M, Malakar CL. Patient decision making: use ofand adherence to telephone-based nurse triage recommendations. . 2002;22:309-317.

ArchPediatr Adolesc Med

7. Baker RC, Schubert CJ, Kirwan KA, Lenkauskas SM, Spaeth JT. After-hourstelephone triage and advice in private and non-private pediatric populations. . 1999;153:292-296.

Arch Pediatr Adolesc Med

8. Crane JD, Benjamin JT. Pediatric residents' telephone triage experience: doparents really follow telephone advice? . 2000;154:71-74.


9. Poole SR, Schmitt BD, Carruth T, Peterson-Smith A, Slusarski M. After-hourstelephone coverage: the application of an area-wide telephone triage and advicesystem for pediatric practices. . 1993;92(5):670-679.

Ann Emerg Med

10. Pane GA, Farner MC, Salness KA. Health care access problems of medicallyindigent emergency department walk-in patients. .1991;20(7):730-733.


11. Rask KJ, Williams MV, Parker RM, McNagny SE. Obstacles predicting lack ofa regular provider and delays in seeking care for patients at an urban public hospital.. 1994;271(24):1931-1933.

Ann Emerg Med

12. Cetta MG, Asplin BR, Fields WW, Yeh CS. Emergency medicine and thedebate over the uninsured: a report from the task force on health care and theuninsured. . 2000;36(3):243-246.