patient satisfaction as a function of emergency department ... · patient satisfaction as a...

THE PRACTICE OF EMERGENCY MEDICINE/ORIGINAL RESEARCH

Patient Satisfaction as a Function of Emergency DepartmentPrevisit Expectations

Ghazwan Toma, MD, MPHWayne Triner, DO, MPHLouise-Ann McNutt, PhD

From the Albany Medical College, Department of Emergency Medicine (Triner), University at Albany,School of Public Health (Toma, Triner, McNutt), Albany, NY.

Study objectives: This study measures the effect of meeting emergency department (ED) patients’ expectationsfor diagnostic and therapeutic interventions on patient satisfaction.

Methods: This was a cross-sectional study of consecutive patients during block enrollment periods surveyed atthe beginning and end of their ED visits. On arrival patients or their surrogates were surveyed about the specificinterventions they expected during their visit. After completion of ED care, they were surveyed about their levelof satisfaction with the entire encounter, assessment of their provider’s interpersonal skills, impression of timespent waiting in the ED, and perceived waiting time. Satisfaction was assessed with categorical responses. Thedegree of concordance of interventions expected and interventions provided was analyzed to determine theireffect on overall ED visit satisfaction.

Results: Nine hundred eighty-seven patients presented during enrollment periods, 821 met inclusion criteria,and complete data were collected on 504 patient encounters. Twenty-nine percent had no previsit expectationsof diagnostic or therapeutic interventions, 24% had a single reported expectation, 47% had multiple interventionexpectations. After adjusting for potential confounders, we could not demonstrate a relationship betweenfulfillment of expectations and satisfaction. We did find a very strong relationship between highly ranked providerinterpersonal skills and ED satisfaction (probability ratio of being “very satisfied” 8.6; 95% confidence interval4.7 to 15.6). Other factors associated with high ED encounter satisfaction were adequate explanations forwaiting times and perception of total time in the ED.

Conclusion: Overall satisfaction was strongly correlated with patient’s assessment of the physician’sinterpersonal skills and was not correlated with whether the physician had met expectations about diagnosticand therapeutic interventions. [Ann Emerg Med. 2009;54:360-367.]

Provide feedback on this article at the journal’s Web site, www.annemergmed.com.

0196-0644/$-see front matterCopyright © 2009 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2009.01.024

INTRODUCTIONBackground

Emergency department (ED) practitioners are responsible forproviding medically appropriate and cost-efficient care with thegoals of patient satisfaction, optimal outcomes, and publichealth benefits. Patient satisfaction has been increasingly used asan outcome measure for health care system performance. EDpatient satisfaction is an incompletely understood concept.Providers may find themselves carrying out activities to enhancesatisfaction without fully knowing whether that activity doesimprove satisfaction. This is particularly problematic if actionstaken to enhance satisfaction conflict with cost efficacy or publichealth responsibilities. Several determinants of ED patientsatisfaction have been explored. Perceived waiting time (asopposed to actual waiting time) has been shown to be a factor inoverall ED satisfaction.1,2 Race and sex interactions are also
related to satisfaction.3,4 Younger patients, black patients, and
360 Annals of Emergency Medicine

those with lower triage acuity express lower satisfaction.5,6

Likewise, language concordance may result in a closerrelationship between patients and providers, positively affectingsatisfaction.7,8 Spahr et al found a relationship between thetreating physician’s awareness of parental expectations andoverall encounter satisfaction.9 They also found a higherpercentage of met expectation when providers were aware of theexpectations. Though not directly explored, their study raisesthe question, did simply meeting parents’ pre-encounterexpectations improve overall satisfaction, or was the awarenessof these expectations enough to initiate other behaviors amongproviders that improved parental satisfaction?

ImportancePhysicians perceive that their patients have specific

expectations during a clinical encounter, and there is a tendencyfor physicians to fulfill these expectations despite medical
appropriateness.10-13 Furthermore, physicians may misperceive
Volume , . : September

http://www.lexisnexis.com/dpartner/process.asp?qs_id=4266

http://www.annemergmed.com

Toma, Triner & McNutt Patient Satisfaction and Previsit Expectations

patient expectations. For example, in a clinic setting withpresumably well-established physician-patient relationships,treating physicians overestimated their patients’ expectation ofantibiotics for upper respiratory infections.14,15 It is poormedical practice to provide unnecessary care, and particularlydysfunctional if mistakenly done so to enhance patientsatisfaction.

Goals of This InvestigationThe goal of this study was to determine the contribution of

meeting a patient’s previsit diagnostic and therapeuticexpectations to overall satisfaction with an ED visit. We alsosought to determine the relative effect of factors that otherinvestigators have shown to influence satisfaction.1-6

MATERIALS AND METHODSStudy Design

We performed a cross-sectional study of patients during theirvisit to an ED. This study was approved by the sponsoringfacility’s institutional review board.

SettingOur study was carried out in the ED of an urban academic

teaching hospital with an emergency medicine residency. ThisED has an annual census of 65,000 patient visits, with an

Editor’s Capsule Summary

What is already known on this topicSome patients come to the emergencydepartment (ED) with specific expectations abouttheir care.

What question this study addressedThis 504-patient, single-ED, observational studyexamined the relationship of patient satisfaction towhether patient expectations were met.

What this study adds to our knowledgeNo important associations between meeting patientexpectations for specific diagnostic and therapeuticinterventions and patient satisfaction were found.As in other studies, satisfaction was associated withpatient ratings of the quality of interpersonal skillsand the time the provider spent with them.

How this might change clinical practiceThis study reinforces that concept that tests andtreatments should be ordered according to medicalnecessity and should not be done simply to meetpatient expectations.

admission rate of 20%. The hospital is a trauma and pediatric


center for a population of 2.4 million urban, suburban, andrural inhabitants.

Selection of ParticipantsThe study population consisted of consecutive patients

treated in the ED during periods of block enrollment betweenJune and September 2006. Blocks were defined by proportionalallocation of discrete areas within the ED, as well as day,evening, week, and weekend periods. We excluded prisoners,the critically ill or injured, those who chose not to participate,and those who were unable to effectively communicate becauseof language. If a patient was unable to provide the necessarydata (children, cognitively impaired, language barrier) and wasaccompanied by someone who could complete the collection ofdata, the person accompanying the patient, if willing, was usedas a surrogate for the patient. We had no occurrences ofrepeated sampling of the same patient on different visits.

We designed our survey to be anonymous and self-administered by the patient or their accompanying surrogate. Ifa respondent or patient required assistance in completing thesurvey, a research associate was available to help. The survey wasdeveloped and piloted during 16 hours of data collection. Theeffectiveness and readability of the survey were determined bythe amount of input required of the research associate andincomplete responses on the survey. There were no formaloutcome measures used for evaluating the instrument duringthe trial, but areas requiring clarification were identified andmodifications were made. Our third draft appeared to bereadable and required little input from the researchers forrespondents to complete. Data collected with the third draft wasincluded in our study analysis because this draft became thefinal survey instrument. Data were collected by one of theauthors (G.T.) and trained research associates. ED providerswere aware that a survey was taking place but were not aware ofthe nature of the survey.

Data Collection and ProcessingFor each patient visit, we collected data in 3 phases: patient

expectations immediately on entry to the ED, patientimpressions measured immediately after completion of care, andretrospective chart review. The initial contact took place beforeany encounter with a physician, nurse practitioner, orphysician’s assistant. During this period, the patient or his orher representative was surveyed to determine demographicfeatures and specific expectations of diagnostic and therapeuticinterventions. Responses were collected categorically (not sure,none, blood test[s], urine test[s], stool test[s], EKG,radiograph[s], ultrasonogram, computed tomography scan,magnetic resonance imaging, or specialty consultation, painmedications, medicine specific to your condition [eg, inhaledbronchodilators], antibiotics, intravenous fluids, hospitaladmission).

Shortly after completion of ED care and before discharge ortransfer to an inpatient unit, participants were surveyed to
determine their overall satisfaction with the ED visit and other
Annals of Emergency Medicine 361

Patient Satisfaction and Previsit Expectations Toma, Triner & McNutt

elements shown to be associated with satisfaction. Theseincluded satisfaction with the diagnostic and therapeuticprocedures done in the ED, physician interpersonal skills,explanation of diagnosis and management, total time spent inthe ED, explanation of the excessive waiting time (if any wasperceived), and the perceived time spent by the provider withthe patient. These responses were collected with a 5-pointcategorical scale (anchored by 1�very satisfied and 5�verydissatisfied). We also asked the patients whether they believedthat they received unnecessary diagnostic or therapeuticinterventions. The survey instrument is available in AppendixE1 (available online at http://www.annemergmed.com).

Finally the ED medical record was reviewed shortly after therespondent left the ED to determine what interventions wereperformed or were scheduled through the ED. If questionsarose, the patient’s provider was approached to verifyinformation from the chart. Other information collected fromthe chart included day of week and shift of the visit, the timespent in the ED from triage to disposition, the final patientdisposition, and diagnostic category (medical or surgical).

The sample size estimation was derived on a bivariateassociation between “expectation(s) met,” and being “highlysatisfied” with the visit. We wanted a sample size sufficient todetect a modest strength of association that the probability ofbeing highly satisfied when expectations were met would be 1.5times higher than when expectations were not met. Ourassumptions for carrying out the sample size calculation werethat the probability of reporting “highly satisfied” with the EDvisit would be 50% in the group with expectations met (P1�.5)and 33% in the group without met expectations (P2�.33).Assigning an � of .05 and seeking a power of 0.8, we calculatedthat we would require 143 patients per group. Anticipatingeffect modification and allowing for variations in percentage ofpatients whose expectations were entirely met, we planned toenroll 500 patients.

Primary Data AnalysisData were entered into an EpiInfo (version 3.5; Centers for

Disease Control and Prevention, Atlanta, GA) and importedinto a SAS database (version 9.0; SAS Institute, Inc., Cary,NC). Our principle outcome measure was the level ofsatisfaction with the ED visit. From review of internal reports,we expected substantial patient satisfaction and thus planned todichotomize satisfaction to “very satisfied” and “other.” Ouranalysis was designed to assess the association betweenexpectations for interventions and overall satisfaction with theED visit. Detailed bivariate analyses were conducted becauseexpectation of interventions was a complex derived variable.First, the association for those expecting each specificintervention (regardless of other expectations) and satisfactionwas computed. Second, the association between those expectingeach specific intervention and satisfaction was computed,stratified by the number of interventions expected (0, 1, 2, 3,�3). Third, for those expecting 2 or 3 interventions, we
reviewed the association between each exact combination of

expectations and satisfaction. Although the numbers becamevery small, we focused on whether there was consistency in thedirection of association or whether the relationship changedwhen different expectation combinations were viewed. We alsolooked for any evidence of effect modification. Although we hadinsufficient power to statistically detect effect modifiers thatwere not very strong, we visually inspected the data to identifyevidence for moderate modification; none was observed.

According to the findings, a summary of 2 perspectives wasdeveloped. First, for those who had only 1 expectation, theassociation between meeting the expectation and satisfaction foreach specific expectation was studied. We then studied theassociation of satisfaction in meeting the expectation for allpatients with any single expectation. Second, we studied theassociation of meeting the proportion expectations met withsatisfaction in persons expressing multiple expectations.

To assess the association between expectations met andoverall ED satisfaction adjusted for potential confounders, wedeveloped a model based on the satisfaction literature.2-8,16

Factors included in the model were proportion of totalexpectations met (100%, 50% to 99.9%, 0.1 to 49.9%, 0%),and the following potential confounders: day and shift (8 AM to4 PM Monday to Friday, 4 PM to 8 AM Monday to Thursday, 4PM Friday to 8 AM Monday), respondent (defined as either thepatient, parent of a minor, or other), patient sex (male, female),age of respondent (18 to 21, 22 to 35, 36 to 50, 51 to 65, �65years), race of respondent (white, black, Hispanic, other), healthinsurance (private insurance, no insurance, Medicare,Medicaid), education of respondent (high school and less, morethan high school), and final patient disposition (admitted,discharged, other). Originally, a log-binomial model wasselected for the analysis because the outcome (patientsatisfaction) is mathematically common and thus the adjustedprevalence ratio must be estimated directly and not with anadjusted odds ratio (as is done with logistic regression). Becausethe log-binomial model did not converge, we analyzed the datawith Poisson regression. The Poisson model with robustvariance estimates also provides adjusted prevalence ratios withreasonable confidence intervals and test statistics.17,18 Ouranalysis of the proportion of expectations met as a predictor ofsatisfaction was carried out only on those expressing at least 1expectation. Significance was defined with an � of 0.05.Interaction terms were not included in the model becausedescriptive assessment of the data did not identify strongcandidates and the sample size was not sufficient to identifymodest effect modification. Model fit focused on the primaryexposure factor of interest (proportion of total expectations met)and influence measures. Each covariate pattern was droppedindividually and the model parameters were estimated todetermine whether any individual or small group of individualshad undue influence on the estimated strength of associationbetween expectations and satisfaction. None were observed. Thesimilarities between estimated unadjusted prevalence ratios and
adjusted prevalence ratios, and similarities between predicted

http://www.annemergmed.com


probabilities and observed probabilities for covariate patternswith substantial numbers, suggested the model reasonably fit thedata. All analysis was carried out with SAS version 9.0.

RESULTSCharacteristics of Study Subjects

There were 987 patients who came to the ED during 343data collection hours; 821 (83%) met inclusion criteria and504 (61%) participated in the study and had completerecords (Figure). The characteristics of the study populationand the presence or absence of expectations are shown inTable 1. Among all the respondents in the study, 29%reported having no expectations, 24% expected 1 diagnosticor therapeutic intervention, 26% expected 2 interventions,11% expected 3 interventions, and 10% expected more than3 interventions in the ED. Hispanics appeared to have hadfewer previsit expectations. Of the 357 expressing at least 1expectation, the most commonly mentioned diagnosticintervention was a blood test, followed by radiographs. Painmedicine was the most commonly expected therapeuticintervention.

The distribution of overall satisfaction scores among the504 respondents was 50% “very satisfied,” 41% “satisfied,”4% “neither satisfied nor dissatisfied,” 3% “dissatisfied,” and2% “very dissatisfied.” Forty-six percent of those who didnot expect any interventions in the ED reported being “verysatisfied,” whereas 51% of those who had expected at least 1procedure were “very satisfied.” For those who expected only1 diagnostic intervention, there was a suggestion of greatersatisfaction when the expectation was met. There was noassociation between having a therapeutic intervention metand patient satisfaction. Although the numbers were smallfor specific expectations, there was nothing to suggestimportant associations (Table 2). There was no associationwith encounter satisfaction or absence of met expectations inthose with multiple expectations. When specificcombinations of expectations were explored (eg, radiographand pain medication), we were unable to appreciate anyassociation between receipt of expected interventions andsatisfaction. However, the numbers involved with each

Figure. Data collection outcome for all eligible patients.

combination were small.


Bivariate analysis of the proportion of expectations met(in those with more than 1 expectation) did not show anassociation with satisfaction. Those provider factorsassociated with being “highly satisfied” included highlyranked interpersonal skills, clarity of explanation of thediagnosis, satisfaction with total ED time, and satisfactionwith the time spent by the physician. On the other hand, therespondent factors of black race, Medicaid as primaryinsurance, admitted patients, and failure to complete highschool were all associated with less satisfaction. The timingof presentation, whether the patient or a surrogate wassurveyed, sex, and age had minimal or no association with

Table 1. Characteristics of respondents, including presence orabsence of diagnostic or therapeutic expectations.

CharacteristicsTotal,

No. (%)

ThoseWithout

Expectations(N�146),No. (%)

Those WithExpectations

(N�358),No. (%)

Day and shift of patient presentation8 AM to 4 PM (Mon. to

Fri.)257 (52) 70 (27) 187 (73)

4 PM to 8 AM (Mon. toThu.)

91 (18) 24 (26) 67 (74)

4 PM Fri. to 8 AM Mon. 149 (30) 50 (34) 99 (66)Missing�7

RespondentPatient 384 (78) 107 (28) 277 (72)Parent 79 (16) 25 (32) 54 (68)Other 31 (6) 11 (35) 20 (64)Missing�10

Sex of the respondentMale 201 (40) 61 (30) 140 (70)Female 296 (60) 83 (28) 213 (72)Missing�7

Age of the respondent, y18–21 46 (9) 15 (33) 31 (67)22–35 193 (39) 62 (32) 131 (68)36–50 153 (31) 38 (25) 115 (75)51–65 70 (14) 20 (29) 50 (71)�65 35 (7) 11 (31) 24 (69)Missing�7

Race of the respondentWhite 298 (60) 93 (31) 205 (69)Black 143 (29) 32 (22) 111 (78)Other 28 (6) 8 (29) 20 (71)Hispanic 28 (6) 13 (46) 15 (54)Missing�7

Primary health insurance of patientMedicaid 145 (29) 42 (29) 103 (71)Medicare 62 (12) 21 (34) 41 (66)Private insurance 228 (46) 60 (26) 168 (74)No insurance 60 (12) 23 (38) 37 (62)Missing�9

Education level of respondentHigh school and less 229 (46) 73 (32) 156 (68)More than high school 266 (54) 71 (27) 195 (73)Missing�9

satisfaction (Table 3).


Table 2. Association of being “very satisfied” stratified on having received expressed previsit expectation(s) of specific diagnostic or therapeutic intervention(s).*

Single expectation, N � 122 One or more expectations, N � 357

N

Number forWhom

ExpectationsWere Met

Very Satisfied

N

Number forWhom

ExpectationsWere Met

Very Satisfied

ExpectationMet, No. (%)

ExpectationNot Met,No. (%) RR (95% CI)

ExpectationMet,

No. (%)

ExpectationNot Met,No. (%) RR (95% CI)

Expected diagnostic interventionsBlood 11 7 3 (43) 2 (50) 0.8 (0.2–3.1) 103 65 33 (50) 17 (5) 1.1 (0.7–1.7)Radiograph 17 12 10 (83) 3 (60) 1.3 (0.6–3) 91 56 34 (60) 17 (48) 1.2 (0.8–1.9)Urine 4 3 1 (33) 0 48 36 16 (44) 7 (58) 0.8 (0.4–1.4)CT scan/magnetic

resonanceimaging

0 0 0 0 33 10 8 (80) 16 (69) 1.1 (0.8–1.7)

EKG 3 3 3 (100) 0 28 22 13 (59) 2 (33) 1.8 (0.5–5.8)Consultation 3 2 1 (50) 0 24 14 6 (43) 4 (40) 1.1 (0.4–2.8)Other (diagnostic) 2 0 0 0 23 6 1 (17) 8 (47) 0.3 (0.1–2.3)Ultrasonography 5 4 2 (50) 0 21 11 4 (36) 5 (50) 0.7 (0.3–2.0)Stool 2 0 0 0 11 2 1 (50) 2 (22) 2.2 (0.4–14.3)Only 1 expectation 47 31 20 (64) 5 (3) 2.1 (0.9–4.5) 246 All expectations

met 11666 (57) 62 (48) 1.2 (0.9–1.5)

246 At least 1expectationmet 164

89 (54) 39 (48) 1.1 (0.9–1.5)

Expected therapeutic interventionsPain medicine 23 19 10 (53) 1 (0.2) 2.1 (0.4–12.2) 152 115 53 (46) 18 (49) 0.9 (0.6–1.4)Specific medicine 22 21 8 (38) 1 (100) 0.4 (0.2–0.7) 80 59 29 (49) 13 (62) 0.8 (0.5–1.2)Antibiotics 20 11 7 (64) 5 (55) 1.1 (0.5–2.4) 78 46 24 (52) 19 (59) 0.9 (0.6–1.3)Intravenous fluid 4 0 0 1 (0.2) 37 10 3 (30) 12 (44) 0.7 (0.2–1.9)Others 5 4 4 (100) 1 (100) 21 12 9 (75) 7 (78) 1.0 (0.6–1.6)Hospital admission 1 1 0 0 14 8 1 (12) 3 (0.5) 0.2 (0.1–1.8)Only 1 expectation 75 56 29 (52) 9 (47) 1.1 (0.6–1.9) 270 All expectations

met 16079 (50) 58 (52.7) 0.9 (0.7–1.2)


100 (48) 37 (60) 0.8 (0.6–1.0)

Diagnostic or therapeutic expected interventionsOnly 1 expectation 122 87 49 (56) 14 (40) 1.4 (0.9–2.2) 357 All expectations

met 15784 (53.5) 100 (50.0) 1.1 (0.9–1.3)


155 (51.5) 29 (51.8) 1 (0.8–1.3)

CT, Computed tomography.*Bivariate analysis of receiving or not receiving specific expectation with being “very satisfied.”

PatientSatisfactionand

PrevisitExpectations

Tom

a,Triner

&M

cNutt

364A

nnalsof

Em

ergencyM

edicineV

olume

,

.

:

September

R

R


Multivariable analysis with Poisson regression also showedthat the proportion of expected diagnostic or therapeuticinterventions received did not improve the probability of arespondent reporting being “very satisfied” with the ED

Table 3. Associations between reporting being “very satisfied”at least 1 expectation of therapeutic or diagnostic intervention.

Variables Total, No. (%)Very Satisfie

No. (%)

Proportion of expectations met, %100 157 (44) 84 (54)50–99.9 114 (32) 58 (51)0.1–49.9 30 (8) 13 (43)0 56 (16) 29 (52)Day and shift

8 AM to 4 PM (Mon. to Fri.) 187 (53) 100 (53)4 PM to 8 AM (Mon. to Thu.) 67 (19) 30 (45)4 PM Fri. to 8 AM Mon. 99 (28) 52 (53)

RespondentPatient 277 (79) 144 (52)Parent of a minor 54 (15) 29 (54)Other 20 (6) 9 (45)

Sex of patientMale 140 (40) 76 (54)Female 213 (60) 106 (50)

Age of respondent, y18–21 3 (9) 13 (42)22–35 131 (37) 68 (52)36–50 115 (33) 60 (52)51–65 50 (14) 26 (52)�65 24 (7) 14 (58)

Race of respondentWhite 205 (58) 120 (58)Black 111 (32) 43 (39)Other 20 (6) 10 (50)Hispanic 15 (4) 7 (47)

Health insurancePrivate insurance 168 (48) 94 (56)No insurance 37 (11) 22 (59)Medicare 41 (12) 24 (59)Medicaid 103 (30) 43 (42)

Education of respondentHigh school and less 156 (44) 72 (46)More than high school 195 (56) 110 (56)

Final patient dispositionDischarged 304 (86) 165 (54)Admitted 42 (12) 14 (33)Other 8 (2) 4 (50)

Satisfaction with interpersonal skills of primary providerVery satisfied 239 (67) 174 (73)Not very satisfied 118 (33) 10 (9)

Satisfaction with explanation of the diagnosisVery satisfied 227 (64) 156 (69)Not very satisfied 128 (36) 27 (21)

espondent’s satisfaction with the total time spent in EDVery satisfied 116 (32) 105 (90)Not very satisfied 240 (67) 79 (33)

espondent’s satisfaction with the provider timeVery satisfied 178 (50) 144 (81)Not very satisfied 179 (50) 40 (22)

encounter (Table 3).


LIMITATIONSBecause we excluded patients deemed too ill to

participate, our findings may not be valid for this group.Similarly, patients with inability to express themselves were

demographic/operational variables among those expressing

Likelihood of Being VerySatisfied, RR (95% CI)

Multivariable Analysis Adjustedfor Demographic and Operational

Characteristics,* RR (95% CI)

Ref Ref0.9 (0.7–1.2) 0.9 (0.7–1.2)0.8 (0.5–1.2) 0.9 (0.6–1.3)

1 (0.7–1.3) 1 (0.7–1.4)

Ref Ref0.8 (0.6–1.1) 0.92 (0.7–1.2)

1 (0.8–1.2) 0.96 (0.8–1.2)

Ref Ref1. (0.8–1.4) 1.1 (0.8–1.4)

0.9 (0.5–1.4) 0.9 (0.6–1.4)

1.1 (0.9–1.3) 1 (0.8–1.3)Ref Ref

0.8 (0.5–1.3) 0.9 (0.6–1.3)Ref Ref

1 (0.8–1.3) 1 (0.8–1.3)1 (0.7–1.4) 1 (0.7–1.3)

1.1 (0.8–1.6) 1 (0.6–1.5)

Ref Ref0.7 (0.5–0.9) 0.8 (0.6–1)0.8 (0.5–1.3) 0.9 (0.6–1.4)0.8 (0.5–1.4) 0.9 (0.5–1.7)

Ref Ref1.1 (0.8–1.4) 1.1 (0.7–1.7)

1 (0.8–1.4) 1.1 (0.8–1.6)0.75 (0.6–0.9) 0.9 (0.7–1.2)

0.82 (0.7–1) 0.8 (0.7–1.1)Ref Ref

Ref Ref0.6 (0.4–0.9) 0.7 (0.5–1.1)0.9 (0.5–1.8) 1 (0.4–2.2)

8.6 (4.7–15.6)Ref

3.3 (2.3–4.6)Ref

2.7 (2.3–3.3)Ref

3.6 (2.7–4.8)Ref

with

d,

also not sampled in our project. This subset of patients


Patient Satisfaction and Previsit Expectations Toma, Triner & McNutt

would generally be expected to report lower scores onsatisfaction.

Measurement of ED patient satisfaction is a complexendeavor. There are no widely accepted and validatedinstruments. Our measure of overall satisfaction was crudeand may not fully capture all nuances of satisfaction.Nevertheless, our surveyed outcome measure resulted infindings similar to those of previous studies, including askewed distribution toward high satisfaction and anassociation with provider interpersonal characteristics.

Although our sample size may be inadequate to detect asmall association between meeting patient expectations fordiagnostic and therapeutic interventions and satisfaction, webelieve that it was adequate to detect an importantassociation.

Last, we did not record provider-level characteristics and thuscould not adjust for clustering of provider demographics. Thislikely resulted in smaller variances than would have beenestimated had clustering been taken into account.

DISCUSSIONThis study provides evidence that although patients often

present to an ED with pre-established diagnostic ortherapeutic expectations for care, the effect of meeting theseexpectations is not associated with overall ED visitsatisfaction. The features that showed meaningful influenceon satisfaction were the respondent’s report that they were“very satisfied” with the provider’s interpersonal skills,receiving an explanation for time spent in the ED, receivingan explanation of the medical condition, and the perceivedtime the physician spent with the patient. The age, race,insurance status, and education levels of the respondent hada demonstrable, but possibly clinically insignificant, effect onsatisfaction.

We found, not surprisingly, that satisfaction is increasedas perceived waiting times are reduced. Interestinglyhowever, estimates of waiting time by patients may not beaccurate.16,19,20 Our study reaffirms the findings thatsatisfaction is linked to waiting times and that adequateexplanation for waits may be rewarded with higher encountersatisfaction.

Education and insurance status, often consideredsurrogates of socioeconomic status, were strongly correlatedwith satisfaction. As those before us have found, markers oflower socioeconomic status were associated with lowersatisfaction scores.21-24 Additionally, ethnic discordancebetween patients and providers has been linked with lowerencounter satisfaction.25 As is the case with most health caredelivery settings, our provider group was heavily skewedtoward nonblack and non-Hispanic ethnicities. Furthermore,our health care providers reside in the middle to uppermiddle socioeconomic strata. Though we did notdemonstrate this, age may also be a factor in determining
level of satisfaction. In our study, those aged between 18 and

21 years demonstrated a trend toward expressing lowersatisfaction with their ED encounter.

The characteristics most associated with high satisfactionwith ED care were highly ranked provider interpersonalskills, explanation of the medical condition, and the lengthof time the provider spent with the patient, each a measureof the effectiveness of communication during the ED visit.Although, by design, we did not measure a provider effect inthis study, differences in communications skills amongproviders likely affect satisfaction. This is important becauseinterpersonal and communication skills can be enhancedthrough experience, training, and feedback.26-30

The findings of this study suggest that investing in thedevelopment of interpersonal and communication skills wouldbe one way to improve patient satisfaction with ED encounters.The act of ordering diagnostic or therapeutic interventions,beyond those indicated for the presenting medical conditions,does not appear to be fruitful in achieving this goal.

Supervising editors: J. Stephan Stapczynski, MD; David L.Schriger, MD, MPH

Author contributions: GT collected data and WT oversaw thecollection. GT, WT, and L-AM were responsible for studydesign. GT and L-AM were responsible for statistical analysis.L-AM was responsible for article review. WT takesresponsibility for the paper as a whole.

Funding and support: By Annals policy, all authors are requiredto disclose any and all commercial, financial, and otherrelationships in any way related to the subject of this articlethat might create any potential conflict of interest. The authorshave stated that no such relationships exist. See theManuscript Submission Agreement in this issue for examplesof specific conflicts covered by this statement.

Publication dates: Received for publication November 3,2007. Revisions received June 10, 2008; July 29, 2008; andOctober 7, 2008. Accepted for publication January 26, 2009.Available online March 12, 2009.

Presented as an abstract at the Society of AcademicEmergency Medicine annual meeting, May 2007, Chicago, IL.

Reprints not available from authors.

Address for correspondence: Wayne Triner, DO, MPH, AlbanyMedical College, Department of Emergency Medicine, 47 NewScotland Avenue (MD-139), Albany, NY 12208; 518-262-3773, fax 518-262-3236; E-mail [email protected].

REFERENCES1. Hedges JR, Trout A, Magnusson AR. Satisfied Patients Exiting the

Emergency Department (SPEED) study. Acad Emerg Med. 2002;9:15-21.

2. Taylor C, Benger JR. Patient satisfaction in emergency medicine.Emerg Med J. 2004;21:528-532.

3. Cooper-Patrick L, Gallo JJ, Gonzales JJ, et al. Race, gender, andpartnership in the patient-physician relationship. JAMA. 1999;
282:583-589.

mailto:[email protected]


4. Derose KP, Hays RD, McCaffrey DF, et al. Does physician genderaffect satisfaction of men and women visiting the emergencydepartment? J Gen Intern Med. 2001;16:218-226.

5. Sun BC, Adams J, Orav EJ, et al. Determinants of patientsatisfaction and willingness to return with emergency care. AnnEmerg Med. 2000;35:426-434.

6. Boudreaux ED, Friedman J, Chansky ME, et al. Emergencydepartment patient satisfaction: examining the role of acuity.Acad Emerg Med. 2004;11:162-168.

7. Mazor SS, Hampers LC, Chande VT, et al. Teaching Spanish topediatric emergency physicians: effects on patient satisfaction.Arch Pediatr Adolesc Med. 2002;156:693-695.

8. Carrasquillo O, Orav EJ, Brennan TA, et al. Impact of languagebarriers on patient satisfaction in an emergency department.J Gen Intern Med. 1999;14:82-87.

9. Spahr CD, Flugstad NA, Brousseau DC. The impact of a briefexpectation survey on parental satisfaction in the pediatricemergency department. Acad Emerg Med. 2006;13:1280-1287.

10. Karras DJ, Ong S, Moran GJ, et al; Emergency ID NET StudyGroup. Antibiotic use for emergency department patients withacute diarrhea: prescribing practices, patient expectations, andpatient satisfaction. Ann Emerg Med. 2003;42:835-842.

11. Macfarlane J, Holmes W, Macfarlane R, et al. Influence ofpatients’ expectations on antibiotic management of acute lowerrespiratory tract illness in general practice: questionnaire study.BMJ. 1997;315:1211-1214.

12. Hart AM, Pepper GA, Gonzales R. Balancing acts: deciding for oragainst antibiotics in acute respiratory infections. J Fam Pract.2006;55:320-325.

13. Briel M, Young J, Tschudi P, et al. Prevalence and influence ofdiagnostic tests for acute respiratory tract infections in primarycare. Swiss Med Wkly. 2006;136:248-253.

14. Linder JA, Singer DE. Desire for antibiotics and antibioticprescribing for adults with upper respiratory tract infections.J Gen Intern Med. 2003;18:795-796.

15. Linder JA, Singer DE, Stafford RS. Association between antibioticprescribing and visit duration in adults with upper respiratory tractinfections. Clin Ther. 2003;25:2419-2430.

16. Thompson DA, Yarnold PR, Williams DR, et al. Effects of actualwaiting time, perceived waiting time, information delivery, andexpressive quality on patient satisfaction in the emergency
department. Ann Emerg Med. 1996;28:657-665.

17. McNutt LA, Wu C, Xue X, et al. Estimating the relative risk incohort studies and clinical trials of common outcomes. Am JEpidemiol. 2003;157:940-943.

18. Spiegelman D, Hertzmark E. Easy SAS calculations for risk orprevalence ratios and differences. Am J Epidemiol. 2005;162:199-200.

19. Thompson DA, Yarnold PR, Adams SL, et al. How accurate arewaiting time perceptions of patients in the emergencydepartment? Ann Emerg Med. 1996;28:652-656.

20. Waseem M, Ravi L, Radeos M, et al. Parental perception ofwaiting time and its influence on parental satisfaction in an urbanpediatric emergency department: are parents accurate indetermining waiting time? South Med J. 2003;96:880-883.

21. Trout A, Magnusson AR, Hedges JR. Patient satisfactioninvestigations and the emergency department: what does theliterature say? Acad Emerg Med. 2000;7:695-709.

22. Cooper LA, Roter DL, Johnson RL, et al. Patient-centeredcommunication, ratings of care, and concordance of patient andphysician race. Ann Intern Med. 2003;139:907-915.

23. Cooper-Patrick L, Gallo JJ, Gonzales JJ, et al. Race, gender, andpartnership in the patient-physician relationship. JAMA.1999;282:583-589.

24. Carrasquillo O, Orav EJ, Brennan TA, et al. Impact of languagebarriers on patient satisfaction in an emergency department.J Gen Intern Med. 1999;2:82-87.

25. Saha S, Komaromy M, Koepsell TD, et al. Patient-physician racialconcordance and the perceived quality and use of health care.Arch Intern Med. 1999;159:997-1004.

26. Hastings A, McKinley RK, Fraser RC. Strengths and weaknessesin the consultation skills of senior medical students:identification, enhancement and curricular change. Med Educ.2006;40:437-443.

27. DiMatteo MR, Hays RD, Prince LM. Relationship of physicians’nonverbal communication skill to patient satisfaction,appointment noncompliance, and physician workload. HealthPsychol. 1986;5:581-594.

28. Stiles WB, Putnam SM, Wolf MH, et al. Interaction exchangestructure and patient satisfaction with medical interviews. MedCare. 1979;17:667-681.

29. Rowland-Morin PA, Carroll JG. Verbal communication skills andpatient satisfaction. A study of doctor-patient interviews. EvalHealth Prof. 1990;13:168-185.

30. Robinson JD, Heritage J. Physicians’ opening questions andpatients’ satisfaction. Patient Educ Couns. 2006;60:279-285.
Epub 2006 Jan 23.

APPENDIX E1. Albany Medical Center patient satisfaction survey.

367.e1 Annals of Emergency Medicine Volume , . : September

Volume , . : September Annals of Emergency Medicine 367.e2

patient satisfaction as a function of emergency department ... · patient satisfaction as a...

Documents