- Email: [email protected]
Determinants of physician reluctance to perform mouth-to-mouth resuscitation
Journal of Clinical Epidemiology 53 (2000) 1054–1061
Determinants of physician reluctance to perform mouth-to-mouth resuscitation Barry E. Brennera,*, David C. Vana, Eliot J. Lazarb, Carlos A. Camargo, Jr.c a
Department of Emergency Medicine, The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, 121 DeKalb Ave., Brooklyn, NY 11201, USA b Department of Internal Medicine, The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, Brooklyn, NY 11201, USA c Department of Emergency Medicine, Massachusetts General Hospital, and Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA Received 29 June 1999; received in revised form 26 January 2000; accepted 16 February 2000
Abstract Objectives: Mouth-to-mouth resuscitation (MMR) is widely taught and promoted. The purpose of this study was to better characterize the observation that health professionals are reluctant to perform MMR and to identify determinants of this reluctance. Methods: 324 residents and faculty at a New York City teaching hospital were anonymously surveyed regarding their reluctance to perform MMR. One year later, medical staff were resurveyed. Results: Reluctance varied across scenarios: 70–80% of physicians were willing to perform MMR on a newborn or child, 40–50% for an unknown man, and 20–30% for a trauma victim or potentially gay man. Physicians reported very similar percentages for each scenario in the two surveys. Factors associated with MMR reluctance were female gender (OR ⫽ 2), resident physician (OR ⫽ 2), and higher perceived risk of contracting HIV from MMR (OR ⫽ 1.4 per unit on 5-point scale). In the year before the survey, 30% of all respondents witnessed an apneic patient who required MMR for whom ventilation was not provided for at least 2 minutes. Conclusions: Many physicians are reluctant to perform MMR. Marked delays in ventilation of apneic patients are occurring. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Cardiac arrest; Cardiopulmonary resuscitation; Human immunodeficiency virus; Acquired immunodeficiency disease; Basic life support
1. Introduction Bystander cardiopulmonary resuscitation (CPR) saves lives with survival depending in part, on whether a bystander witnessed the cardiac arrest and whether CPR was actually performed [1,2]. This clear-cut benefit has led public health officials to advocate lay person CPR training and to encourage health care workers to function as leaders, trainers, and advocates of the bystander CPR program. Mouth-tomouth resuscitation (MMR) is an important part of the program and of basic life support in general [3–5], although its effect on survival has not been studied independently. MMR requires no additional equipment and permits timely ventilation, the primary mode of intervention in victims of opiate drug ingestion, drowning, and electrical injury [5]. In children, in whom respiratory causes of arrest predominate, early MMR can reverse the underlying hypoxia and increase the chances of successful resuscitation [5]. * Corresponding author. Tel.: 718-250-8733; fax: 718-250-6528. E-mail address: [email protected] (B.E. Brenner)
In recent years, several studies have reported a reluctance to perform MMR by physicians and other health care personnel, both as a citizen bystanders in the community and in the course of their in-hospital duties [6–8]. Despite evidence that transmission of human immunodeficiency virus (HIV) is an unlikely complication of MMR [9], medical professionals often cite this fear as an important reason for their reluctance [6–8]. There remain concerns, however, about the reliability and validity of this research on reluctance to perform MMR. The purpose of the present study was to better characterize the observation that health professionals are reluctant to perform MMR and to identify determinants of this reluctance. Specifically, does reluctance to perform MMR vary between different medical specialties and are the medical providers in a given specialty more likely to provide MMR for their specific patients than other patients. For example, are pediatricians more likely to provide MMR for children and obstetrics-gynecology for neonates? Because anesthesiology is a medical specialty emphasizing the importance of airway management, we suspected
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(00)00 2 3 0 - 4
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
that personnel in this field would be more willing to perform MMR. We sought to address a variety of methodologic issues about this finding of reluctance to perform MMR and to determine if prolonged delays in ventilation actually are occurring. 2. Methods This study was conducted at a 642-bed New York City teaching hospital with residencies in anesthesia, emergency medicine, family practice, internal medicine, obstetricsgynecology, pediatrics, and surgery. All house staff were certified in Basic Life Support within the 2 years before the study. All categorical residents (n ⫽ 227, 96% of total) and full-time faculty (n ⫽ 96, 77% of total) were anonymously surveyed regarding their age, sex, training status, and their reluctance to perform MMR in six cardiac arrest scenarios (Table 1). Respondents were asked to indicate which action they would initiate: (1) MMR and external chest compressions; (2) MMR only; (3) external chest compressions only; and (4) no active intervention, other than summoning for help. To ensure that respondents would understand the need and appropriateness for full CPR and eliminate the hypothetical option of reliance on others, respondents were instructed to presume that they witnessed the arrest onset, that all victims were pulseless and apneic, that no other capable bystanders were present, and that no barrier mask was available. Each of the six scenarios featured a single victim of cardiac arrest and was designed to imply variable risk of infection of communicable disease. Surveys were handed out at the time of regular hospital conferences, and, after the form was completed, the respondent would check his/her name off a list and place the form in a stack of completed forms. To achieve a high response rate, the forms were
1055
handed out at multiple regularly recurring conferences over 6 weeks. To examine the reliability and validity of this anonymous self-administered survey, a second survey was administered 1 year later in the same fashion but only over 1 week at the initial regularly recurring conference. A total of 225 medical staff responded to the same scenarios reported in Table 1; the group was comprised of 113 original respondents, 111 new respondents, and 1 respondent of unknown status. Because some respondents might refuse to take any risks, no matter how small, the second survey also included new questions to assess how reluctance to perform MMR varied with perceived risk of disease transmission (such as HIV) and perceived likelihood of successful CPR outcome. Second, respondents were asked if they agreed with the statement: “However low the risks for disease transmission or high the chances for a successful outcome from CPR, I will not accept any risk associated with performing MMR on strangers (scenario 1)” Third, we asked the respondent to order the sequence of the resuscitations if he/she were presented with all six scenarios simultaneously and presuming that there were no infectious disease risks. Respondents were instructed that each resuscitation would take 15 minutes, and, by the time you arrived to resuscitate the next victim, that the chance of survival of the subsequent victim had been decreased. Victims that were attended to first or second were considered of high priority. Fourth, we wanted to assess the frequency of actual observations of delayed ventilation by physicians; to this end, we asked if within the last year the respondent had witnessed an apneic patient requiring MMR for whom ventilation was not provided for at least 2 minutes. Lastly, we sought to determine the consistency of our findings with prior studies in the literature. Recent articles on health professionals’ reluctance to perform
Table 1 Cardiac arrest scenarios for the assessment of willingness to perform mouth-to-mouth resuscitation 1. Unknown man You are in a neighborhood grocery store when a man of average dress and build, about 40 years old, who you have never seen before collapses before you. You summon for help. Then would you: (a) Peform mouth-to-mouth resuscitation (MMR) and external chest compressions (b) Perform MMR only; no external chest compressions (c) Perform external chest compressions only; no MMR (d) Not perform MMR or external chest compressions. 2. Trauma You are in an urban area when you see a car collide with a tree. You immediately run to the scene. In the car is a healthy appearing, well-dressed 35-year-old man with a slight amount of blood on his face. After summoning for help, would you . . . [see four options above]. 3. Child It is Halloween, and a 6-year-old girl and her 6-year-old friend knock at your door. The child points to her friend and says, “Something is wrong with Mary.” Mary then collapses in front of you. You summon for help. Then, would you . . .[see four options above]. 4. Elderly person You are visiting a nursing home on your off time, when a nurse becomes upset because a 70-year-old functional patient she’d given medicines to 5 minutes earlier “looks terrible.” After you summon for help, would you . . . [see four options above]. 5. Newborn You are at home, and a taxi pulls up on the street. The driver yells, “Hey, this woman’s having a baby.” You run over and deliver a normally developed, fullterm baby. You cut the cord. The baby does not turn pink or respond to agitation. It is not breathing and is pulseless. The mother says she saw her ob-gyn doctor yesterday, and “everything was fine.” After summoning help, would you . . . [see four options above]. 6. Gay neighborhood You are in a predominantly gay neighborhood on a Sunday in a trendy bookstore buying a newspaper when a well-dressed 26-year-old man clutches his chest and falls to the ground. You summon help. Would you . . . [see four options above].
1056
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
MMR were identified with a MEDLINE search that used the following title search terms: resuscitation, arrest, helping, and CPR. These articles were supplemented by retrieval of relevant citations and included in a qualitative review. 2.1. Statistical analysis All analyses were performed using Stata 5.0 (StataCorp, College Station, TX). Data are presented as proportions or means [with standard deviation (SD)]. The association between MMR reluctance and other factors was examined using chi-squared test, Student’s t-test, Wilcoxon rank sum test, and non-parametric test for trend, as appropriate. Variables associated with MMR reluctance at a P ⬍ 0.2 were evaluated for inclusion in multivariate logistic regression models. All odds ratios (OR) are presented with 95% confidence intervals (CI). Respondents were instructed that if they probably would perform MMR to answer in the affirmative and if they probably would not perform MMR to answer in the negative; percentages are reported as willingness to perform MMR. All P-values are two-sided, with P ⬍ 0.05 considered statistically significant.
Table 2 Characteristics of respondents Characteristic Age, yrs (mean ⫾ SD) Male (%) Race/ethnicity (%) White Black Hispanic Asian Other Hospital department (%) Anesthesia Emergency medicine Family practice Internal medicine Obstetrics-gynecology Pediatrics Surgery Status (%) PGY1 PGY2 PGY3 PGY4 or more Faculty
Survey 1 (n ⫽ 324)
Survey 2 (n ⫽ 225)
35 ⫾ 8 65
34 ⫾ 7 69
43 13 5 29 9
42 17 5 24 13
7 13 5 38 9 16 13
10 16 10 35 6 12 10
26 20 18 6 30
25 18 16 11 29
PGY denotes post-graduate year.
3. Results 3.1. Respondents The overall response rate was 90% to the primary survey, and 62% to the survey 1 year later. The increased response to the primary survey was due to the extra 5 weeks in obtaining more surveys responses from different respondents at each of the regularly recurring conferences. Respondents to survey 1 were quite similar to those who completed survey 2 (Table 2). The average age was 35 years, about two-thirds were men, about half were in the departments of internal medicine and pediatrics, and two-thirds were residents in the first 3 years of their training. One respondent did not report his training status. Survey 2 provided overall findings that were similar to those from the survey 1, so most of the following results are based on the initial survey. 3.2. Overall and by specialty Reluctance varied across scenarios (Table 3). In general, more physicians were willing to perform MMR on children (neonate, 67%; child, 79%) than adults (trauma, 23%; unknown, 43%) (P ⬍ 0.001). Physicians were less willing to provide MMR to victims with more perceived high risk for transmission of infectious diseases (trauma and gay neighborhood) as compared to scenarios involving unknown risks (elderly person and unknown man) (P ⬍ 0.001). The overall willingness to provide MMR was lower for the neonate (67%) vs the child [79% (P ⬍ 0.001)], and this may have been due to respondents’ perception of greater risk in providing MMR to a neonate than child. As previously mentioned, across the two surveys physicians reported very sim-
ilar willingness to perform MMR. For example, the overall willingness in the second survey was 48% for the unknown man, 27% for the trauma victim, 79% for the child, 44% for the elderly person, 69% for the neonate, and 27% for the gay neighborhood. Specialty was not consistently associated with reluctance to perform MMR (Table 3). Anesthesia, whose specialty is airway management, was no different from other specialties in their willingness to perform MMR in the various scenarios (all P ⬎ 0.17). Physicians also were not more willing to provide MMR to the patients that they regularly treat than were members of any other specialty group. For example, physicians in ob-gyn or pediatrics were equally reluctant to provide MMR for a newborn or child as were physicians in other specialties (both P ⬎ 0.60). Most of the physicians that were unwilling to perform MMR would provide external compression and, in general, only 10–38% of the respondents would not assist the victim at all by performing any external compression or MMR. Anesthesia was more likely to forgo CPR entirely than were the other specialities. For example, in the child scenario, 27% of anesthesiologists would not perform any MMR or ECC as compared to 6% of all the other respondents (P ⫽ 0.001). 3.3. Gender and medical experience In every scenario, men were more likely than women to perform MMR. Men were 20% more likely to perform MMR in the unknown scenario (51% males vs. 30% females in the unknown scenario; P ⬍ 0.001) and 10% more likely in the trauma and gay scenarios (26% vs. 16% will-
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
1057
Table 3 Willingness to perform mouth-to-mouth resuscitation (MMR), overall and by medical specialtya Percentage of respondents willing to perform MMR Specialty
n
Unknownb
Traumac
Childd
Elderlye
Newbornf
Gayg
Total Anesthesia Emergency medicine Family practice Internal medicine Ob-gyn Pediatrics Surgery
324 22 42 15 122 29 53 41
43 55 50 40 34 35 53 51
23 29 21 33 19 21 25 27
79 68 81 93 79 76 83 78
40 27 45 60 33 38 55 34
67 73 74 73 63 66 72 63
22 19 16 27 22 17 23 20
a
Respondents willing to perform MMR and external compression, or MMR only, were regarded as willing to perform MMR. Unknown man from scenario 1 in Table 1. c Trauma from scenario 2 in Table 1. d Child from scenario 3 in Table 1. e Elderly person from scenario 4 in Table 1. f Newborn from scenario 5 in Table 1. g Gay neighborhood from scenario 6 in Table 1. b
ingness in both scenarios; both P ⫽ 0.04). The reluctance of females to perform MMR was true not only where the risks were unclear or perceived to be high, but even where the risks were low, such as in the elderly scenario (45% vs. 30% willingness; p ⫽ 0.007). The increased willingness of men to perform MMR was not significant in the child and neonatal scenarios, where women’s willingness was only 4–5% lower (both P ⬎ 0.24). In each scenario, as medical experience increased during PGY1 to PGY3 training, willingness to perform MMR progressively decreased. For example, the unknown man scenario, 47% of the PGY1 residents were willing to perform MMR, 39% for PGY2, 29% for PGY3, and 25% for PGY4 or more (P for trend ⫽ 0.01). However, in every scenario but the elderly person, the faculty were more willing to provide MMR than the residents. For example, in the unknown man scenario, 38% of residents were willing to perform MMR vs. 54% of faculty (P ⫽ 0.009). 3.4. Validity There was no difference in the percentage of respondents willing to perform MMR on the unknown man among those who rated the chance of CPR success as high (16–50%) versus those who thought the man had a worse prognosis (54% vs. 44%, P ⫽ 0.14). (Results in this validity section are all based on the second survey (see Methods).) Among those respondents who perceived that the risk of contracting HIV was low (⬍0.05%) vs. those who perceived the risk to be higher, there was greater willingness to perform MMR (58% vs. 36%; P ⫽ 0.001). Overall, 82% of respondents felt that the risk of acquiring HIV from the unknown man was 0.5% or less. However, 55% (95% CI, 48– 62%) of respondents were unwilling to accept any risk in performing MMR on strangers. In the multiple victim scenario, children had the highest priority for resuscitation (92%), followed by neonates (84%). The trauma, unknown, and gay neighborhood vic-
tims were intermediate, and the elderly victim was given the lowest priority by 98% of the respondents. Overall, 30% (95% CI, 25–35%) of the respondents had witnessed an apneic victim that required MMR but did not receive ventilation for over 2 minutes. This observation was noted in all specialties: 24% for anesthesia, 24% for emergency medicine, 53% for family practice, 38% for internal medicine, 17% for pediatrics, 17% for obstetric-gynecology, and 33% for surgery. 3.5. Multivariate analysis To further examine the association between physician factors and reluctance to perform MMR, we constructed multivariate logistic regression models (Table 4). Both surveys showed that female gender were independent determinants of MMR reluctance. Resident status also tended to be an independent determinant of MMR reluctance in the first survey [P ⫽ 0.02 (first survey) and P ⫽ 0.06 (second survey)]. The second survey permitted simultaneous control for potential reasons and demonstrated the expected relations for perceived risk of contracting HIV (P ⫽ 0.009) and perceived chance of CPR success (P ⫽ 0.06); HIV risk appeared to be the stronger of these two independent determinants, because the 95% CI for perceived chance for CPR success was 0.6–1.0. Even after controlling for these two potential reasons, female gender and resident status were associated with increased reluctance to perform MMR.
4. Discussion This study demonstrates that many physicians are reluctant to perform MMR and that the magnitude of this reluctance varies across scenarios: only 70–80% of physicians were willing to perform MMR on a newborn or child, 40– 50% for an unknown man, and 20–30% for a trauma victim or potentially gay man. The suggested delays (or frank de-
1058
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
Table 4 Multivariate models of determinants of physician reluctance to perform mouth-to-mouth resuscitation Survey 1
Survey 2
Factor
OR
95% CI
P-value
OR
95% CI
P-value
Age (per 10 years) Sex (1 ⫽ male, 2 ⫽ female) Black (0 ⫽ no, 1 ⫽ yes) Resident (0 ⫽ no, 1 ⫽ yes) Perceived risk of HIV (5-point scale)a Perceived success of CPR (5-point scale)a
1.1 2.5 1.0 2.1 – –
0.8–1.6 1.5–4.2 0.5–2.0 1.1–4.0 – –
0.58 ⬍0.001 0.93 0.02 – –
1.1 2.1 0.5 2.3 1.4 0.8
0.6–2.0 1.1–4.3 0.2–1.2 1.0–5.3 1.1–1.9 0.6–1.0
0.67 0.03 0.11 0.06 0.009 0.06
Abbreviations: OR, odds ratio; CI, confidence interval; HIV, human immunodeficiency virus; CPR, cardiopulmonary resuscitation. a These factors are only available in survey 2. Higher scores indicate higher risk and higher success, respectively.
nial of potentially life-saving assistance) appears to result from fear of HIV transmission, rather than skepticism about the efficacy of the maneuver. These findings have profound public health implications. Several studies have reported similar findings (Table 5), although many had low survey response rates or focused on a specific subgroup (e.g., residents, gay men, or internists). In general, between 39 to 70% of respondents have been willing to perform MMR on a stranger. The willingness of internists in the present study to perform MMR was consistent with that from prior studies [6,7,10,11], although there appeared to be a downward trend over time (56%, 54%, 50%, 39%, and 34% in studies 2, 5, 6, 11, and the present study). The surveys have come from a variety of locales, including communities with both low [10,12,13,15,17,19] and high [6,7,8,11,14,16,18] HIV prevalence. Paramedics in New York were unusual in that they reported they would refuse entirely to perform MMR on a stranger, although their willingness to perform under other circumstances was increased [8]. The gay males were unique in their high willingness to perform MMR on a stranger or, for that matter, in a variety of cardiac arrest victims [14]. This willingness was consistently decreased if the gay respondent were HIV-positive. This uniquely high willingness may be due to the intensive educational efforts to which the gay community has been subjected concerning the risk of HIV transmission and the negligible risk of transmission by the oral route [9]. However, physicians and out-of-hospital care providers also have been intensively educated about the routes of HIV transmission and the associated risks. Furthermore, in situations which involved a potential romantic liaison with an unknown partner, in which the vast majority of physicians and out-of-hospital providers would engage in oral contact, almost none would perform MMR if that partner suffered a cardiopulmonary arrest [6,8]. It is likely that the difference and high degree of willingness of the male gay community to perform MMR involves the fact that this group has grown accustomed to the infectious disease risks associated with their lifestyle and apply that “comfort level” to cardiac arrest situations. In the present study, there was no easily identifiable specialty that was more reluctant (or willing) to perform MMR. The type of patient that the specialty treats also did not in-
fluence the respondent’s willingness to perform MMR. For example, obstetrician-gynecologists and pediatricians were no more likely to perform MMR on a child or neonate than internists or surgeons. By contrast, female gender was associated with decreased willingness to perform MMR in every scenario. Women’s increased reluctance to perform MMR was even true for scenarios where potential risk of infectious disease transmission was low, such as the elderly person (scenario 4). This finding has been noted in earlier studies [10–12]. Reduced helping behavior toward strangers experiencing bleeding emergencies has been demonstrated among nonmedical women, as well [21]. In addition to gender, training level was related to willingness to provide MMR in the first survey. There was a progressive decrease in willingness with each year of house staff training but the faculty, in general, were more willing to perform MMR than the residents. Possible explanations for these findings have been house staff sensory overload, sleep deprivation, or lack of emotional arousal by a cardiac arrest, all of which would be cumulative over the years of resident training and less operative in faculty [11]. In almost all studies to date, fear of infectious diseases, particularly HIV, has been the main reason for MMR reluctance (Table 5). We posited that perceived risks for acquiring HIV, and perceived chances for successful resuscitation, both might affect willingness to perform MMR. Our data suggest, however, that only the perceived risk for acquiring HIV during the resuscitation was related to willingness to perform MMR. Although we found that over 80% of the respondents did not vastly overestimate the actual risk, and felt that the chance of acquiring HIV from the unknown man was less than 0.5%, it appears that the quantitative degree of risk of HIV transmission during MMR is not very important. For the majority of respondents (55%), they are unwilling to take any risk, no matter how small, even if the chances of successful resuscitation are high. In the multivariate modeling, risk of HIV transmission emerged as an independent determinant of physician reluctance to perform MMR. One of the major criticisms of the literature to date (Table 5) has been the absence of reliability and validity testing. At the most basic level, could responses to the fictitious
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
1059
Table 5 Studies on willingness to perform mouth-to-mouth resuscitation on unknown victim
Author 1. Ornato et al.[13]
2. Brenner et al.[6]
3. Brenner[14]
4. Mandel[15]
5. Brenner et al.[7]
6. Brenner et al.[10]
7. Melanson et al.[16]
8. Horowitz[17]
9. Locke et al[12]
10. Hew et al.[8]
11. Brenner et al.[11]
12. Gross et al.[18]
13. Lester et al.[19]
14. Pepe et al.[23] 15. Present study
Sample, (methoda) location CPR instructors (mail) Virginia Internists Nurse (mail) Los Angeles Gay men (interview) Los Angeles Lay public (interview) Seattle Medical resident (interview) Los Angeles Medical residents (interview) Loma Linda, CA Pre-hospital care (group)d Pennsylvania Health workers (group) Sacramento, CA Heart Center list (mail) Tucson, AZ EMTse (interview) Paramedics Residents Applicants (group) Brooklyn, NY Internists (group) New York, NY CPR trainees (90% mail, 10% interview) Cardiff, UK Laypersons Pittsburgh, PA Physicians (group) Brooklyn
% Response
n
% Willingb
Comment
Reason
Yearc
AIDS
1988
Infectious diseases
1991
1794
41
29
433 152
73 80
56 20
89% in health care, 40% actually delayed MMR, 40% witnessed MMR delays No MMR for romantic partner met at a dance
200
94 94
93 85
% willing if subject HIV (⫺) % willing if subject HIV (⫹)
– –
1991
–
45
Gender of stranger not listed
–
1992
74 82
100 99
54 45
Out-of-hospital willingness In-hospital willingness
HIV
1993
58
100 100
43 50
HIV
1994
40
80
50
In-hospital willingness Out-of-hospital willingness 63% walked away or did only ECC in actual arrests Drowned child
HIV
1994
61
Out-of-hospital unknown man In-hospital, no HIV/hepatitis risk In-hospital, unknown HIV/hepatitis risk 80% lay; 20% health care providers
–
1994–95
Infectious diseases
1995
388
379
975
30
47 66 33 55
77 27
100 –
43 0
40% walked away or did only ECC in actual arrests
Infectious disease
1996
99
39 70
–
–
1996
1996
1998
– –
97 169
–
54f
96 96
54 69
Airplane scenario Restaurant scenario
368
46
82
Laypersons
Reluctant MMR Not responsible Infectious disease –
476
45
68
Population based
–
1998
324 225
90 62
48 43
All residents and attendings in consecutive years
HIV risk
1996–97
Abbreviations: CPR, cardiopulmonary resuscitation; MMR, mouth-to-mouth resuscitation; AIDS, acquired immune deficiency syndrome; HIV, human immunodeficiency virus; ECC, external chest compression; EMT, emergency medical technician. a Method by which survey was performed. Mail means the survey was sent to respondent by mail; group means the survey was done in an group setting (such as a conference); and interview means that the survey was completed by an interviewer asking the respondent the questions, either in person or by telephone. b Percent willing to perform MMR on an unknown victim. c Year the study was performed. d Group refers to the written survey being distributed and collected during a meeting (e.g., lecture, orientation). e Emergency medical technicians. f Willingness to perform CPR.
1060
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
scenarios not reflect what the respondent would actually do when confronted with a cardiac arrest victim? We designed the present study to carefully address many aspects of this methodologic concern. Taking each of the items in turn: 1. Is there any reason to believe that nonresponders to the survey were more willing to perform MMR? In the first survey, only 43% of the respondents would perform MMR in the unknown man scenario. There were only 36 of 360 potential respondents that were missed in this survey. Even if each of these nonrespondents had agreed to perform MMR, the percent willing to perform MMR would increase to only 49% (95% CI, 44–54%). 2. Are the results reliable? That is, are they similar on re-testing? The two surveys, repeated 1 year apart, provided very similar results. 3. Are the results from this one hospital consistent with the findings of other researchers? Several other studies have found comparable reluctance to provide MMR to strangers among a variety of groups (Table 5). The cited studies are from several geographically different settings and suggest that our findings are not unique to this urban hospital. Although the size of the study cities is different, there are no surveys in rural settings and data from clearly suburban populations are limited. 4. Does the data collection approach affect the results? The present study and those in Table 5 comprise 14 studies (and 17 respondent groups) with information about willingness to perform MMR on a stranger. The list can be classified into three approaches: interview, survey in group setting, and survey received through the mail. The respondents were grouped by the manner in which the survey was performed and weighted according to the number of respondents in each group. Each method of survey completion resulted in similar willingness to perform MMR in an unknown victim (mailed, 40%; group setting, 47%; interview, 43%). 5. Might the respondent not understand the question? Might the respondent not know the answer to the question being asked and therefore provide a meaningless response? Among this group of physicians, it seems highly unlikely that respondents had not considered MMR and their possible response to a situation in which they would be expected to perform this potentially life-saving intervention. Moreover, respondents that perceived higher risk for the transmission of HIV were more reluctant to perform MMR than those with lower perceptions of risk. 6. Might the respondent perceive that the survey was not completely anonymous and therefore alter his/her survey? This problem is unlikely because the forms were unmarked and reviewed in a completely anonymous fashion. Moreover, we believe that people would tend to overstate their willingness to perform MMR because
performing MMR would be the more socially acceptable response. If so, then the true percent willingness to perform MMR on a stranger would actually be lower. 7. Ultimately, we return to the most basic question: Do the answers correspond to what they are intended to measure? In other words, a respondent may write on the survey that he/she would not perform MMR, but when confronted with an actual arrest he/she may be willing to perform MMR, “risking his/her life” rather than watching helplessly as a person dies. It is impossible to test the respondent’s behavior in actual arrest situations. However reluctance to perform MMR in actual cardiac arrests does occur. In a series of 38 consecutive non-intensive care unit cardiac or respiratory arrests in adults, 63% of cardiac arrest patients were witnessed to be apneic and receive no ventilation for over 1 minute and 18% were witnessed to be apneic and receive no ventilation for over 3 minutes [22]. These delays in ventilation were occurring on the inpatient wards, while the staff was searching for the bag-valve-mask to provide ventilation. In our study, 30% of the respondents witnessed a patient apneic for over 2 minutes in the last year. Although these delays in ventilation are dependent on recall of time, these findings are consistent with other studies [8,10,13], where 40–63% of respondents either performed only chest compressions or walked away from the actual victims of arrest rather than perform MMR. Notwithstanding the popularity of bystander-initiated CPR in cities such as Seattle [20], the willingness among the lay population to perform CPR, even if instructed by a trained telephone dispatcher is not clear. In New York City, less than 25% of lay persons will perform CPR during direction by a telephone dispatcher (personal communication, Lorraine Giordano, Director of Emergency Medical Services, New York City, 1996). Thus, the survey results appear to reflect a reluctance to perform MMR in actual cardiopulmonary arrests. In conclusion, we found that a disturbingly high percentage of physicians are reluctant to perform MMR, a potentially life-saving intervention. The independent determinants of this reluctance include female gender, resident physician, and concern about transmission of HIV. We examined a variety of potential problems with this survey finding and conclude that our results are both valid and reliable. Indeed, we document that marked delays in ventilation of apneic patients are perceived to be occurring. The problem appears to stem from physicians’ unwillingness to accept any risk of HIV transmission so it does not appear that additional education about HIV transmission will help. Furthermore, our data suggest that improvements in CPR efficacy also will not eliminate this unfortunate situation because chances of CPR success were not a consistent factor in the decisionmaking process. A lack of commitment to perform MMR by such a diverse group of New York City physicians, and by
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
other health professionals from a variety of locales, does not bode well for the future of MMR and its effects on the performance of bystander CPR.
[9]
[10]
Acknowledgments Dr. Camargo is supported by grant HL-03533 from the National Institutes of Health (Bethesda, MD). This study was approved by the hospital’s Institutional Review Board.
[11]
References
[13]
[1] Stiell IG, Wells GA, DeMaio VJ, Spaite DW, Field BJ III, Munkley DP, Lyver MB, Luinstra LG, Ward R. Modifiable factors associated with improved cardiac arrest survival in a multicenter basic life support/defibrillation system: OPALS Study Phase I results. Ontario Prehospital Advanced Life Support. Ann Emerg Med 1999;33:44–50. [2] Stiell IG, Wells GA, Field BJ III, Spaite DW, DeMaio VJ, Ward R, Munkley DP, Lyver MB, Luinstra LG, Campeau T, Maloney J, Dagnone F. Improved out-of-hospital cardiac arrest survival through the inexpensive optimization of an existing defibrillation program: OPALS study phase II. Ontario Prehospital Advanced Life Support. JAMA 1998;281:1175–81. [3] Elam JO, Brown ES, Elder JD. Artificial respiration by mouth-to-mouth method: a study of the respiratory gas exchange of paralyzed patients ventilated by operator’s expired air. N Engl J Med 1954;250:749–54. [4] Safar P, Escarraga LA, Elam JO. A comparison of the mouth-tomouth and mouth-to-airway methods of artificial respiration with the chest-pressure arm-lift methods. N Engl J Med 1958;258:671–7. [5] Emergency Cardiac Care Committee and Subcommittees, American Heart Association. Guidelines for cardiopulmonary resuscitation and emergency cardiac care. I: Introduction. JAMA 1992;268:2172–83. [6] Brenner BE, Kauffman J. Reluctance of internists and medical nurses to perform mouth-to-mouth resuscitation. Arch Intern Med 1993;153: 1763–9. [7] Brenner B, Stark B, Kauffman J. The reluctance of house staff to perform mouth-to-mouth resuscitation in the inpatient setting: what are the considerations? Resuscitation 1994;28:185–93. [8] Hew P, Brenner B, Kauffman J. Reluctance of paramedics and emer-
[12]
[14] [15]
[16] [17] [18]
[19]
[20]
[21]
[22] [23]
1061
gency medical technicians to perform mouth-to-mouth resuscitation. J Emerg Med 1997; 15:279–84. Mejicano GC, Maki DG. Infections acquired during cardiopulmonary resuscitation: estimating the risk and defining strategies for prevention. Ann Intern Med 1998;129:813–28. Brenner B, Kauffman J, Sachter JJ. Comparison of the reluctance of house staff of metropolitan and suburban hospitals to perform mouthto-mouth resuscitation. Resuscitation 1996;32:5–12. Brenner BE, Van DC, Cheng D, Lazar E. Determinants of reluctance to perform CPR among residents and applicants: the impact of experience on helping behavior. Resuscitation 1997;35:203–11. Locke CJ, Berg R, Sanders A, Davis MF, Milander MM, Kern KB, Ewy GA. Bystander cardiopulmonary resuscitation: concerns about mouth-to-mouth contact. Arch Intern Med 1995;155:938–43. Ornato JP, Hallagan LF, McMahan SB, Peeples EH, Rostafinski AG. Attitudes of BCLS instructors about mouth-to-mouth resuscitation during the AIDS epidemic. Ann Emerg Med 1990;19:151–6. Brenner B. Willingness of male homosexuals to perform mouth-tomouth resuscitation. Resuscitation 1994;27:23–30. Mandel L. Attitudes and willingness to perform CPR in Seattle. Presented at the American Hearth Association, National Conference on Cardiopulmonary Resuscitation and Emergency Cardiac Care, Dallas, TX; 1992. Melanson S, Pronchik D. Mouth-to-mouth resuscitation by EMS providers. Prehosp Dis Med 1995;10:137. Horowitz BZ, Matheny L. Health care professionals’ willingness to do mouth-to-mouth resuscitation. West J Med 1997;167:392–7. Gross CP, Resiman AB, Schwartz MD. The physician as ambivalent samaritan: will internists resuscitate victims of out-of-hospital emergencies? J Gen Intern Med 1998;13:491–4. Lester C, Donnelly P, Assar D. Lay CPR trainees four years on: a postal study of re-training, confidence, and willingness to attempt to resuscitation. Resuscitation 1998;37:S51. Culley LL, Clark JJ, Eisenberg MS, Larsen MP. Dispatcher-assisted telephone CPR: common delays and time standards for delivery. Ann Emerg Med 1991;20:362–6. Shotland RL, Heinold WD. Bystander response to arterial bleeding: helping skills, the decision-making process and differentiating the helping response. J Pers Soc Psychol 1985;49:347–56. Brenner BE, Kauffmann J. Response to cardiac arrests in a hospital setting: delays in ventilation. Resuscitation 1996;31:17–23. Pepe PE, Scheatzle MD, Forjuoh SN, Dearwater SR, Aronson AA. Would more people perform basic CPR if mouth-to-mouth ventilation was optional. Acad Emerg Med 1999;6:430.
Determinants of physician reluctance to perform mouth-to-mouth resuscitation Barry E. Brennera,*, David C. Vana, Eliot J. Lazarb, Carlos A. Camargo, Jr.c a
Department of Emergency Medicine, The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, 121 DeKalb Ave., Brooklyn, NY 11201, USA b Department of Internal Medicine, The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, Brooklyn, NY 11201, USA c Department of Emergency Medicine, Massachusetts General Hospital, and Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA Received 29 June 1999; received in revised form 26 January 2000; accepted 16 February 2000
Abstract Objectives: Mouth-to-mouth resuscitation (MMR) is widely taught and promoted. The purpose of this study was to better characterize the observation that health professionals are reluctant to perform MMR and to identify determinants of this reluctance. Methods: 324 residents and faculty at a New York City teaching hospital were anonymously surveyed regarding their reluctance to perform MMR. One year later, medical staff were resurveyed. Results: Reluctance varied across scenarios: 70–80% of physicians were willing to perform MMR on a newborn or child, 40–50% for an unknown man, and 20–30% for a trauma victim or potentially gay man. Physicians reported very similar percentages for each scenario in the two surveys. Factors associated with MMR reluctance were female gender (OR ⫽ 2), resident physician (OR ⫽ 2), and higher perceived risk of contracting HIV from MMR (OR ⫽ 1.4 per unit on 5-point scale). In the year before the survey, 30% of all respondents witnessed an apneic patient who required MMR for whom ventilation was not provided for at least 2 minutes. Conclusions: Many physicians are reluctant to perform MMR. Marked delays in ventilation of apneic patients are occurring. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Cardiac arrest; Cardiopulmonary resuscitation; Human immunodeficiency virus; Acquired immunodeficiency disease; Basic life support
1. Introduction Bystander cardiopulmonary resuscitation (CPR) saves lives with survival depending in part, on whether a bystander witnessed the cardiac arrest and whether CPR was actually performed [1,2]. This clear-cut benefit has led public health officials to advocate lay person CPR training and to encourage health care workers to function as leaders, trainers, and advocates of the bystander CPR program. Mouth-tomouth resuscitation (MMR) is an important part of the program and of basic life support in general [3–5], although its effect on survival has not been studied independently. MMR requires no additional equipment and permits timely ventilation, the primary mode of intervention in victims of opiate drug ingestion, drowning, and electrical injury [5]. In children, in whom respiratory causes of arrest predominate, early MMR can reverse the underlying hypoxia and increase the chances of successful resuscitation [5]. * Corresponding author. Tel.: 718-250-8733; fax: 718-250-6528. E-mail address: [email protected] (B.E. Brenner)
In recent years, several studies have reported a reluctance to perform MMR by physicians and other health care personnel, both as a citizen bystanders in the community and in the course of their in-hospital duties [6–8]. Despite evidence that transmission of human immunodeficiency virus (HIV) is an unlikely complication of MMR [9], medical professionals often cite this fear as an important reason for their reluctance [6–8]. There remain concerns, however, about the reliability and validity of this research on reluctance to perform MMR. The purpose of the present study was to better characterize the observation that health professionals are reluctant to perform MMR and to identify determinants of this reluctance. Specifically, does reluctance to perform MMR vary between different medical specialties and are the medical providers in a given specialty more likely to provide MMR for their specific patients than other patients. For example, are pediatricians more likely to provide MMR for children and obstetrics-gynecology for neonates? Because anesthesiology is a medical specialty emphasizing the importance of airway management, we suspected
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(00)00 2 3 0 - 4
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
that personnel in this field would be more willing to perform MMR. We sought to address a variety of methodologic issues about this finding of reluctance to perform MMR and to determine if prolonged delays in ventilation actually are occurring. 2. Methods This study was conducted at a 642-bed New York City teaching hospital with residencies in anesthesia, emergency medicine, family practice, internal medicine, obstetricsgynecology, pediatrics, and surgery. All house staff were certified in Basic Life Support within the 2 years before the study. All categorical residents (n ⫽ 227, 96% of total) and full-time faculty (n ⫽ 96, 77% of total) were anonymously surveyed regarding their age, sex, training status, and their reluctance to perform MMR in six cardiac arrest scenarios (Table 1). Respondents were asked to indicate which action they would initiate: (1) MMR and external chest compressions; (2) MMR only; (3) external chest compressions only; and (4) no active intervention, other than summoning for help. To ensure that respondents would understand the need and appropriateness for full CPR and eliminate the hypothetical option of reliance on others, respondents were instructed to presume that they witnessed the arrest onset, that all victims were pulseless and apneic, that no other capable bystanders were present, and that no barrier mask was available. Each of the six scenarios featured a single victim of cardiac arrest and was designed to imply variable risk of infection of communicable disease. Surveys were handed out at the time of regular hospital conferences, and, after the form was completed, the respondent would check his/her name off a list and place the form in a stack of completed forms. To achieve a high response rate, the forms were
1055
handed out at multiple regularly recurring conferences over 6 weeks. To examine the reliability and validity of this anonymous self-administered survey, a second survey was administered 1 year later in the same fashion but only over 1 week at the initial regularly recurring conference. A total of 225 medical staff responded to the same scenarios reported in Table 1; the group was comprised of 113 original respondents, 111 new respondents, and 1 respondent of unknown status. Because some respondents might refuse to take any risks, no matter how small, the second survey also included new questions to assess how reluctance to perform MMR varied with perceived risk of disease transmission (such as HIV) and perceived likelihood of successful CPR outcome. Second, respondents were asked if they agreed with the statement: “However low the risks for disease transmission or high the chances for a successful outcome from CPR, I will not accept any risk associated with performing MMR on strangers (scenario 1)” Third, we asked the respondent to order the sequence of the resuscitations if he/she were presented with all six scenarios simultaneously and presuming that there were no infectious disease risks. Respondents were instructed that each resuscitation would take 15 minutes, and, by the time you arrived to resuscitate the next victim, that the chance of survival of the subsequent victim had been decreased. Victims that were attended to first or second were considered of high priority. Fourth, we wanted to assess the frequency of actual observations of delayed ventilation by physicians; to this end, we asked if within the last year the respondent had witnessed an apneic patient requiring MMR for whom ventilation was not provided for at least 2 minutes. Lastly, we sought to determine the consistency of our findings with prior studies in the literature. Recent articles on health professionals’ reluctance to perform
Table 1 Cardiac arrest scenarios for the assessment of willingness to perform mouth-to-mouth resuscitation 1. Unknown man You are in a neighborhood grocery store when a man of average dress and build, about 40 years old, who you have never seen before collapses before you. You summon for help. Then would you: (a) Peform mouth-to-mouth resuscitation (MMR) and external chest compressions (b) Perform MMR only; no external chest compressions (c) Perform external chest compressions only; no MMR (d) Not perform MMR or external chest compressions. 2. Trauma You are in an urban area when you see a car collide with a tree. You immediately run to the scene. In the car is a healthy appearing, well-dressed 35-year-old man with a slight amount of blood on his face. After summoning for help, would you . . . [see four options above]. 3. Child It is Halloween, and a 6-year-old girl and her 6-year-old friend knock at your door. The child points to her friend and says, “Something is wrong with Mary.” Mary then collapses in front of you. You summon for help. Then, would you . . .[see four options above]. 4. Elderly person You are visiting a nursing home on your off time, when a nurse becomes upset because a 70-year-old functional patient she’d given medicines to 5 minutes earlier “looks terrible.” After you summon for help, would you . . . [see four options above]. 5. Newborn You are at home, and a taxi pulls up on the street. The driver yells, “Hey, this woman’s having a baby.” You run over and deliver a normally developed, fullterm baby. You cut the cord. The baby does not turn pink or respond to agitation. It is not breathing and is pulseless. The mother says she saw her ob-gyn doctor yesterday, and “everything was fine.” After summoning help, would you . . . [see four options above]. 6. Gay neighborhood You are in a predominantly gay neighborhood on a Sunday in a trendy bookstore buying a newspaper when a well-dressed 26-year-old man clutches his chest and falls to the ground. You summon help. Would you . . . [see four options above].
1056
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
MMR were identified with a MEDLINE search that used the following title search terms: resuscitation, arrest, helping, and CPR. These articles were supplemented by retrieval of relevant citations and included in a qualitative review. 2.1. Statistical analysis All analyses were performed using Stata 5.0 (StataCorp, College Station, TX). Data are presented as proportions or means [with standard deviation (SD)]. The association between MMR reluctance and other factors was examined using chi-squared test, Student’s t-test, Wilcoxon rank sum test, and non-parametric test for trend, as appropriate. Variables associated with MMR reluctance at a P ⬍ 0.2 were evaluated for inclusion in multivariate logistic regression models. All odds ratios (OR) are presented with 95% confidence intervals (CI). Respondents were instructed that if they probably would perform MMR to answer in the affirmative and if they probably would not perform MMR to answer in the negative; percentages are reported as willingness to perform MMR. All P-values are two-sided, with P ⬍ 0.05 considered statistically significant.
Table 2 Characteristics of respondents Characteristic Age, yrs (mean ⫾ SD) Male (%) Race/ethnicity (%) White Black Hispanic Asian Other Hospital department (%) Anesthesia Emergency medicine Family practice Internal medicine Obstetrics-gynecology Pediatrics Surgery Status (%) PGY1 PGY2 PGY3 PGY4 or more Faculty
Survey 1 (n ⫽ 324)
Survey 2 (n ⫽ 225)
35 ⫾ 8 65
34 ⫾ 7 69
43 13 5 29 9
42 17 5 24 13
7 13 5 38 9 16 13
10 16 10 35 6 12 10
26 20 18 6 30
25 18 16 11 29
PGY denotes post-graduate year.
3. Results 3.1. Respondents The overall response rate was 90% to the primary survey, and 62% to the survey 1 year later. The increased response to the primary survey was due to the extra 5 weeks in obtaining more surveys responses from different respondents at each of the regularly recurring conferences. Respondents to survey 1 were quite similar to those who completed survey 2 (Table 2). The average age was 35 years, about two-thirds were men, about half were in the departments of internal medicine and pediatrics, and two-thirds were residents in the first 3 years of their training. One respondent did not report his training status. Survey 2 provided overall findings that were similar to those from the survey 1, so most of the following results are based on the initial survey. 3.2. Overall and by specialty Reluctance varied across scenarios (Table 3). In general, more physicians were willing to perform MMR on children (neonate, 67%; child, 79%) than adults (trauma, 23%; unknown, 43%) (P ⬍ 0.001). Physicians were less willing to provide MMR to victims with more perceived high risk for transmission of infectious diseases (trauma and gay neighborhood) as compared to scenarios involving unknown risks (elderly person and unknown man) (P ⬍ 0.001). The overall willingness to provide MMR was lower for the neonate (67%) vs the child [79% (P ⬍ 0.001)], and this may have been due to respondents’ perception of greater risk in providing MMR to a neonate than child. As previously mentioned, across the two surveys physicians reported very sim-
ilar willingness to perform MMR. For example, the overall willingness in the second survey was 48% for the unknown man, 27% for the trauma victim, 79% for the child, 44% for the elderly person, 69% for the neonate, and 27% for the gay neighborhood. Specialty was not consistently associated with reluctance to perform MMR (Table 3). Anesthesia, whose specialty is airway management, was no different from other specialties in their willingness to perform MMR in the various scenarios (all P ⬎ 0.17). Physicians also were not more willing to provide MMR to the patients that they regularly treat than were members of any other specialty group. For example, physicians in ob-gyn or pediatrics were equally reluctant to provide MMR for a newborn or child as were physicians in other specialties (both P ⬎ 0.60). Most of the physicians that were unwilling to perform MMR would provide external compression and, in general, only 10–38% of the respondents would not assist the victim at all by performing any external compression or MMR. Anesthesia was more likely to forgo CPR entirely than were the other specialities. For example, in the child scenario, 27% of anesthesiologists would not perform any MMR or ECC as compared to 6% of all the other respondents (P ⫽ 0.001). 3.3. Gender and medical experience In every scenario, men were more likely than women to perform MMR. Men were 20% more likely to perform MMR in the unknown scenario (51% males vs. 30% females in the unknown scenario; P ⬍ 0.001) and 10% more likely in the trauma and gay scenarios (26% vs. 16% will-
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
1057
Table 3 Willingness to perform mouth-to-mouth resuscitation (MMR), overall and by medical specialtya Percentage of respondents willing to perform MMR Specialty
n
Unknownb
Traumac
Childd
Elderlye
Newbornf
Gayg
Total Anesthesia Emergency medicine Family practice Internal medicine Ob-gyn Pediatrics Surgery
324 22 42 15 122 29 53 41
43 55 50 40 34 35 53 51
23 29 21 33 19 21 25 27
79 68 81 93 79 76 83 78
40 27 45 60 33 38 55 34
67 73 74 73 63 66 72 63
22 19 16 27 22 17 23 20
a
Respondents willing to perform MMR and external compression, or MMR only, were regarded as willing to perform MMR. Unknown man from scenario 1 in Table 1. c Trauma from scenario 2 in Table 1. d Child from scenario 3 in Table 1. e Elderly person from scenario 4 in Table 1. f Newborn from scenario 5 in Table 1. g Gay neighborhood from scenario 6 in Table 1. b
ingness in both scenarios; both P ⫽ 0.04). The reluctance of females to perform MMR was true not only where the risks were unclear or perceived to be high, but even where the risks were low, such as in the elderly scenario (45% vs. 30% willingness; p ⫽ 0.007). The increased willingness of men to perform MMR was not significant in the child and neonatal scenarios, where women’s willingness was only 4–5% lower (both P ⬎ 0.24). In each scenario, as medical experience increased during PGY1 to PGY3 training, willingness to perform MMR progressively decreased. For example, the unknown man scenario, 47% of the PGY1 residents were willing to perform MMR, 39% for PGY2, 29% for PGY3, and 25% for PGY4 or more (P for trend ⫽ 0.01). However, in every scenario but the elderly person, the faculty were more willing to provide MMR than the residents. For example, in the unknown man scenario, 38% of residents were willing to perform MMR vs. 54% of faculty (P ⫽ 0.009). 3.4. Validity There was no difference in the percentage of respondents willing to perform MMR on the unknown man among those who rated the chance of CPR success as high (16–50%) versus those who thought the man had a worse prognosis (54% vs. 44%, P ⫽ 0.14). (Results in this validity section are all based on the second survey (see Methods).) Among those respondents who perceived that the risk of contracting HIV was low (⬍0.05%) vs. those who perceived the risk to be higher, there was greater willingness to perform MMR (58% vs. 36%; P ⫽ 0.001). Overall, 82% of respondents felt that the risk of acquiring HIV from the unknown man was 0.5% or less. However, 55% (95% CI, 48– 62%) of respondents were unwilling to accept any risk in performing MMR on strangers. In the multiple victim scenario, children had the highest priority for resuscitation (92%), followed by neonates (84%). The trauma, unknown, and gay neighborhood vic-
tims were intermediate, and the elderly victim was given the lowest priority by 98% of the respondents. Overall, 30% (95% CI, 25–35%) of the respondents had witnessed an apneic victim that required MMR but did not receive ventilation for over 2 minutes. This observation was noted in all specialties: 24% for anesthesia, 24% for emergency medicine, 53% for family practice, 38% for internal medicine, 17% for pediatrics, 17% for obstetric-gynecology, and 33% for surgery. 3.5. Multivariate analysis To further examine the association between physician factors and reluctance to perform MMR, we constructed multivariate logistic regression models (Table 4). Both surveys showed that female gender were independent determinants of MMR reluctance. Resident status also tended to be an independent determinant of MMR reluctance in the first survey [P ⫽ 0.02 (first survey) and P ⫽ 0.06 (second survey)]. The second survey permitted simultaneous control for potential reasons and demonstrated the expected relations for perceived risk of contracting HIV (P ⫽ 0.009) and perceived chance of CPR success (P ⫽ 0.06); HIV risk appeared to be the stronger of these two independent determinants, because the 95% CI for perceived chance for CPR success was 0.6–1.0. Even after controlling for these two potential reasons, female gender and resident status were associated with increased reluctance to perform MMR.
4. Discussion This study demonstrates that many physicians are reluctant to perform MMR and that the magnitude of this reluctance varies across scenarios: only 70–80% of physicians were willing to perform MMR on a newborn or child, 40– 50% for an unknown man, and 20–30% for a trauma victim or potentially gay man. The suggested delays (or frank de-
1058
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
Table 4 Multivariate models of determinants of physician reluctance to perform mouth-to-mouth resuscitation Survey 1
Survey 2
Factor
OR
95% CI
P-value
OR
95% CI
P-value
Age (per 10 years) Sex (1 ⫽ male, 2 ⫽ female) Black (0 ⫽ no, 1 ⫽ yes) Resident (0 ⫽ no, 1 ⫽ yes) Perceived risk of HIV (5-point scale)a Perceived success of CPR (5-point scale)a
1.1 2.5 1.0 2.1 – –
0.8–1.6 1.5–4.2 0.5–2.0 1.1–4.0 – –
0.58 ⬍0.001 0.93 0.02 – –
1.1 2.1 0.5 2.3 1.4 0.8
0.6–2.0 1.1–4.3 0.2–1.2 1.0–5.3 1.1–1.9 0.6–1.0
0.67 0.03 0.11 0.06 0.009 0.06
Abbreviations: OR, odds ratio; CI, confidence interval; HIV, human immunodeficiency virus; CPR, cardiopulmonary resuscitation. a These factors are only available in survey 2. Higher scores indicate higher risk and higher success, respectively.
nial of potentially life-saving assistance) appears to result from fear of HIV transmission, rather than skepticism about the efficacy of the maneuver. These findings have profound public health implications. Several studies have reported similar findings (Table 5), although many had low survey response rates or focused on a specific subgroup (e.g., residents, gay men, or internists). In general, between 39 to 70% of respondents have been willing to perform MMR on a stranger. The willingness of internists in the present study to perform MMR was consistent with that from prior studies [6,7,10,11], although there appeared to be a downward trend over time (56%, 54%, 50%, 39%, and 34% in studies 2, 5, 6, 11, and the present study). The surveys have come from a variety of locales, including communities with both low [10,12,13,15,17,19] and high [6,7,8,11,14,16,18] HIV prevalence. Paramedics in New York were unusual in that they reported they would refuse entirely to perform MMR on a stranger, although their willingness to perform under other circumstances was increased [8]. The gay males were unique in their high willingness to perform MMR on a stranger or, for that matter, in a variety of cardiac arrest victims [14]. This willingness was consistently decreased if the gay respondent were HIV-positive. This uniquely high willingness may be due to the intensive educational efforts to which the gay community has been subjected concerning the risk of HIV transmission and the negligible risk of transmission by the oral route [9]. However, physicians and out-of-hospital care providers also have been intensively educated about the routes of HIV transmission and the associated risks. Furthermore, in situations which involved a potential romantic liaison with an unknown partner, in which the vast majority of physicians and out-of-hospital providers would engage in oral contact, almost none would perform MMR if that partner suffered a cardiopulmonary arrest [6,8]. It is likely that the difference and high degree of willingness of the male gay community to perform MMR involves the fact that this group has grown accustomed to the infectious disease risks associated with their lifestyle and apply that “comfort level” to cardiac arrest situations. In the present study, there was no easily identifiable specialty that was more reluctant (or willing) to perform MMR. The type of patient that the specialty treats also did not in-
fluence the respondent’s willingness to perform MMR. For example, obstetrician-gynecologists and pediatricians were no more likely to perform MMR on a child or neonate than internists or surgeons. By contrast, female gender was associated with decreased willingness to perform MMR in every scenario. Women’s increased reluctance to perform MMR was even true for scenarios where potential risk of infectious disease transmission was low, such as the elderly person (scenario 4). This finding has been noted in earlier studies [10–12]. Reduced helping behavior toward strangers experiencing bleeding emergencies has been demonstrated among nonmedical women, as well [21]. In addition to gender, training level was related to willingness to provide MMR in the first survey. There was a progressive decrease in willingness with each year of house staff training but the faculty, in general, were more willing to perform MMR than the residents. Possible explanations for these findings have been house staff sensory overload, sleep deprivation, or lack of emotional arousal by a cardiac arrest, all of which would be cumulative over the years of resident training and less operative in faculty [11]. In almost all studies to date, fear of infectious diseases, particularly HIV, has been the main reason for MMR reluctance (Table 5). We posited that perceived risks for acquiring HIV, and perceived chances for successful resuscitation, both might affect willingness to perform MMR. Our data suggest, however, that only the perceived risk for acquiring HIV during the resuscitation was related to willingness to perform MMR. Although we found that over 80% of the respondents did not vastly overestimate the actual risk, and felt that the chance of acquiring HIV from the unknown man was less than 0.5%, it appears that the quantitative degree of risk of HIV transmission during MMR is not very important. For the majority of respondents (55%), they are unwilling to take any risk, no matter how small, even if the chances of successful resuscitation are high. In the multivariate modeling, risk of HIV transmission emerged as an independent determinant of physician reluctance to perform MMR. One of the major criticisms of the literature to date (Table 5) has been the absence of reliability and validity testing. At the most basic level, could responses to the fictitious
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
1059
Table 5 Studies on willingness to perform mouth-to-mouth resuscitation on unknown victim
Author 1. Ornato et al.[13]
2. Brenner et al.[6]
3. Brenner[14]
4. Mandel[15]
5. Brenner et al.[7]
6. Brenner et al.[10]
7. Melanson et al.[16]
8. Horowitz[17]
9. Locke et al[12]
10. Hew et al.[8]
11. Brenner et al.[11]
12. Gross et al.[18]
13. Lester et al.[19]
14. Pepe et al.[23] 15. Present study
Sample, (methoda) location CPR instructors (mail) Virginia Internists Nurse (mail) Los Angeles Gay men (interview) Los Angeles Lay public (interview) Seattle Medical resident (interview) Los Angeles Medical residents (interview) Loma Linda, CA Pre-hospital care (group)d Pennsylvania Health workers (group) Sacramento, CA Heart Center list (mail) Tucson, AZ EMTse (interview) Paramedics Residents Applicants (group) Brooklyn, NY Internists (group) New York, NY CPR trainees (90% mail, 10% interview) Cardiff, UK Laypersons Pittsburgh, PA Physicians (group) Brooklyn
% Response
n
% Willingb
Comment
Reason
Yearc
AIDS
1988
Infectious diseases
1991
1794
41
29
433 152
73 80
56 20
89% in health care, 40% actually delayed MMR, 40% witnessed MMR delays No MMR for romantic partner met at a dance
200
94 94
93 85
% willing if subject HIV (⫺) % willing if subject HIV (⫹)
– –
1991
–
45
Gender of stranger not listed
–
1992
74 82
100 99
54 45
Out-of-hospital willingness In-hospital willingness
HIV
1993
58
100 100
43 50
HIV
1994
40
80
50
In-hospital willingness Out-of-hospital willingness 63% walked away or did only ECC in actual arrests Drowned child
HIV
1994
61
Out-of-hospital unknown man In-hospital, no HIV/hepatitis risk In-hospital, unknown HIV/hepatitis risk 80% lay; 20% health care providers
–
1994–95
Infectious diseases
1995
388
379
975
30
47 66 33 55
77 27
100 –
43 0
40% walked away or did only ECC in actual arrests
Infectious disease
1996
99
39 70
–
–
1996
1996
1998
– –
97 169
–
54f
96 96
54 69
Airplane scenario Restaurant scenario
368
46
82
Laypersons
Reluctant MMR Not responsible Infectious disease –
476
45
68
Population based
–
1998
324 225
90 62
48 43
All residents and attendings in consecutive years
HIV risk
1996–97
Abbreviations: CPR, cardiopulmonary resuscitation; MMR, mouth-to-mouth resuscitation; AIDS, acquired immune deficiency syndrome; HIV, human immunodeficiency virus; ECC, external chest compression; EMT, emergency medical technician. a Method by which survey was performed. Mail means the survey was sent to respondent by mail; group means the survey was done in an group setting (such as a conference); and interview means that the survey was completed by an interviewer asking the respondent the questions, either in person or by telephone. b Percent willing to perform MMR on an unknown victim. c Year the study was performed. d Group refers to the written survey being distributed and collected during a meeting (e.g., lecture, orientation). e Emergency medical technicians. f Willingness to perform CPR.
1060
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
scenarios not reflect what the respondent would actually do when confronted with a cardiac arrest victim? We designed the present study to carefully address many aspects of this methodologic concern. Taking each of the items in turn: 1. Is there any reason to believe that nonresponders to the survey were more willing to perform MMR? In the first survey, only 43% of the respondents would perform MMR in the unknown man scenario. There were only 36 of 360 potential respondents that were missed in this survey. Even if each of these nonrespondents had agreed to perform MMR, the percent willing to perform MMR would increase to only 49% (95% CI, 44–54%). 2. Are the results reliable? That is, are they similar on re-testing? The two surveys, repeated 1 year apart, provided very similar results. 3. Are the results from this one hospital consistent with the findings of other researchers? Several other studies have found comparable reluctance to provide MMR to strangers among a variety of groups (Table 5). The cited studies are from several geographically different settings and suggest that our findings are not unique to this urban hospital. Although the size of the study cities is different, there are no surveys in rural settings and data from clearly suburban populations are limited. 4. Does the data collection approach affect the results? The present study and those in Table 5 comprise 14 studies (and 17 respondent groups) with information about willingness to perform MMR on a stranger. The list can be classified into three approaches: interview, survey in group setting, and survey received through the mail. The respondents were grouped by the manner in which the survey was performed and weighted according to the number of respondents in each group. Each method of survey completion resulted in similar willingness to perform MMR in an unknown victim (mailed, 40%; group setting, 47%; interview, 43%). 5. Might the respondent not understand the question? Might the respondent not know the answer to the question being asked and therefore provide a meaningless response? Among this group of physicians, it seems highly unlikely that respondents had not considered MMR and their possible response to a situation in which they would be expected to perform this potentially life-saving intervention. Moreover, respondents that perceived higher risk for the transmission of HIV were more reluctant to perform MMR than those with lower perceptions of risk. 6. Might the respondent perceive that the survey was not completely anonymous and therefore alter his/her survey? This problem is unlikely because the forms were unmarked and reviewed in a completely anonymous fashion. Moreover, we believe that people would tend to overstate their willingness to perform MMR because
performing MMR would be the more socially acceptable response. If so, then the true percent willingness to perform MMR on a stranger would actually be lower. 7. Ultimately, we return to the most basic question: Do the answers correspond to what they are intended to measure? In other words, a respondent may write on the survey that he/she would not perform MMR, but when confronted with an actual arrest he/she may be willing to perform MMR, “risking his/her life” rather than watching helplessly as a person dies. It is impossible to test the respondent’s behavior in actual arrest situations. However reluctance to perform MMR in actual cardiac arrests does occur. In a series of 38 consecutive non-intensive care unit cardiac or respiratory arrests in adults, 63% of cardiac arrest patients were witnessed to be apneic and receive no ventilation for over 1 minute and 18% were witnessed to be apneic and receive no ventilation for over 3 minutes [22]. These delays in ventilation were occurring on the inpatient wards, while the staff was searching for the bag-valve-mask to provide ventilation. In our study, 30% of the respondents witnessed a patient apneic for over 2 minutes in the last year. Although these delays in ventilation are dependent on recall of time, these findings are consistent with other studies [8,10,13], where 40–63% of respondents either performed only chest compressions or walked away from the actual victims of arrest rather than perform MMR. Notwithstanding the popularity of bystander-initiated CPR in cities such as Seattle [20], the willingness among the lay population to perform CPR, even if instructed by a trained telephone dispatcher is not clear. In New York City, less than 25% of lay persons will perform CPR during direction by a telephone dispatcher (personal communication, Lorraine Giordano, Director of Emergency Medical Services, New York City, 1996). Thus, the survey results appear to reflect a reluctance to perform MMR in actual cardiopulmonary arrests. In conclusion, we found that a disturbingly high percentage of physicians are reluctant to perform MMR, a potentially life-saving intervention. The independent determinants of this reluctance include female gender, resident physician, and concern about transmission of HIV. We examined a variety of potential problems with this survey finding and conclude that our results are both valid and reliable. Indeed, we document that marked delays in ventilation of apneic patients are perceived to be occurring. The problem appears to stem from physicians’ unwillingness to accept any risk of HIV transmission so it does not appear that additional education about HIV transmission will help. Furthermore, our data suggest that improvements in CPR efficacy also will not eliminate this unfortunate situation because chances of CPR success were not a consistent factor in the decisionmaking process. A lack of commitment to perform MMR by such a diverse group of New York City physicians, and by
B.E. Brenner et al. / Journal of Clinical Epidemiology 53 (2000) 1054–1061
other health professionals from a variety of locales, does not bode well for the future of MMR and its effects on the performance of bystander CPR.
[9]
[10]
Acknowledgments Dr. Camargo is supported by grant HL-03533 from the National Institutes of Health (Bethesda, MD). This study was approved by the hospital’s Institutional Review Board.
[11]
References
[13]
[1] Stiell IG, Wells GA, DeMaio VJ, Spaite DW, Field BJ III, Munkley DP, Lyver MB, Luinstra LG, Ward R. Modifiable factors associated with improved cardiac arrest survival in a multicenter basic life support/defibrillation system: OPALS Study Phase I results. Ontario Prehospital Advanced Life Support. Ann Emerg Med 1999;33:44–50. [2] Stiell IG, Wells GA, Field BJ III, Spaite DW, DeMaio VJ, Ward R, Munkley DP, Lyver MB, Luinstra LG, Campeau T, Maloney J, Dagnone F. Improved out-of-hospital cardiac arrest survival through the inexpensive optimization of an existing defibrillation program: OPALS study phase II. Ontario Prehospital Advanced Life Support. JAMA 1998;281:1175–81. [3] Elam JO, Brown ES, Elder JD. Artificial respiration by mouth-to-mouth method: a study of the respiratory gas exchange of paralyzed patients ventilated by operator’s expired air. N Engl J Med 1954;250:749–54. [4] Safar P, Escarraga LA, Elam JO. A comparison of the mouth-tomouth and mouth-to-airway methods of artificial respiration with the chest-pressure arm-lift methods. N Engl J Med 1958;258:671–7. [5] Emergency Cardiac Care Committee and Subcommittees, American Heart Association. Guidelines for cardiopulmonary resuscitation and emergency cardiac care. I: Introduction. JAMA 1992;268:2172–83. [6] Brenner BE, Kauffman J. Reluctance of internists and medical nurses to perform mouth-to-mouth resuscitation. Arch Intern Med 1993;153: 1763–9. [7] Brenner B, Stark B, Kauffman J. The reluctance of house staff to perform mouth-to-mouth resuscitation in the inpatient setting: what are the considerations? Resuscitation 1994;28:185–93. [8] Hew P, Brenner B, Kauffman J. Reluctance of paramedics and emer-
[12]
[14] [15]
[16] [17] [18]
[19]
[20]
[21]
[22] [23]
1061
gency medical technicians to perform mouth-to-mouth resuscitation. J Emerg Med 1997; 15:279–84. Mejicano GC, Maki DG. Infections acquired during cardiopulmonary resuscitation: estimating the risk and defining strategies for prevention. Ann Intern Med 1998;129:813–28. Brenner B, Kauffman J, Sachter JJ. Comparison of the reluctance of house staff of metropolitan and suburban hospitals to perform mouthto-mouth resuscitation. Resuscitation 1996;32:5–12. Brenner BE, Van DC, Cheng D, Lazar E. Determinants of reluctance to perform CPR among residents and applicants: the impact of experience on helping behavior. Resuscitation 1997;35:203–11. Locke CJ, Berg R, Sanders A, Davis MF, Milander MM, Kern KB, Ewy GA. Bystander cardiopulmonary resuscitation: concerns about mouth-to-mouth contact. Arch Intern Med 1995;155:938–43. Ornato JP, Hallagan LF, McMahan SB, Peeples EH, Rostafinski AG. Attitudes of BCLS instructors about mouth-to-mouth resuscitation during the AIDS epidemic. Ann Emerg Med 1990;19:151–6. Brenner B. Willingness of male homosexuals to perform mouth-tomouth resuscitation. Resuscitation 1994;27:23–30. Mandel L. Attitudes and willingness to perform CPR in Seattle. Presented at the American Hearth Association, National Conference on Cardiopulmonary Resuscitation and Emergency Cardiac Care, Dallas, TX; 1992. Melanson S, Pronchik D. Mouth-to-mouth resuscitation by EMS providers. Prehosp Dis Med 1995;10:137. Horowitz BZ, Matheny L. Health care professionals’ willingness to do mouth-to-mouth resuscitation. West J Med 1997;167:392–7. Gross CP, Resiman AB, Schwartz MD. The physician as ambivalent samaritan: will internists resuscitate victims of out-of-hospital emergencies? J Gen Intern Med 1998;13:491–4. Lester C, Donnelly P, Assar D. Lay CPR trainees four years on: a postal study of re-training, confidence, and willingness to attempt to resuscitation. Resuscitation 1998;37:S51. Culley LL, Clark JJ, Eisenberg MS, Larsen MP. Dispatcher-assisted telephone CPR: common delays and time standards for delivery. Ann Emerg Med 1991;20:362–6. Shotland RL, Heinold WD. Bystander response to arterial bleeding: helping skills, the decision-making process and differentiating the helping response. J Pers Soc Psychol 1985;49:347–56. Brenner BE, Kauffmann J. Response to cardiac arrests in a hospital setting: delays in ventilation. Resuscitation 1996;31:17–23. Pepe PE, Scheatzle MD, Forjuoh SN, Dearwater SR, Aronson AA. Would more people perform basic CPR if mouth-to-mouth ventilation was optional. Acad Emerg Med 1999;6:430.