Original Contribution

Physicians’ impression on the elders’ functionality influences decision making for emergency care

Alejandro Rodriguez-Molinero MD?, Maria Lopez-Dieguez MD, Ana I. Tabuenca MSc, Juan J. de la Cruz MSc, Jose R. Banegas MD, PhD

Department of Preventive Medicine and Public Health. Universidad Autonoma de Madrid, CIBERESP, 28029 Madrid, Spain

Received 15 November 2008; revised 17 March 2009; accepted 19 March 2009

Abstract

Background and aims: This study analyzes the elements that compose the emergency physicians’ criterion for selecting elderly patients for intensive care treatment. This issue has not been studied in-depth. Methods: A cross-sectional study was conducted at 4 University teaching hospitals, covering 101 randomly selected elderly patients admitted to emergency department and their respective physicians.

Physicians were asked to forecast their plans for treatment or therapeutic abstention, in the event that patients might require aggressive measures (cardiopulmonary resuscitation or admission to critical care units). Data were collected on physicians’ reasons for taking such decisions and their patients’ functional capacity and cognitive status (Katz index and Informant Questionnaire on cognitive decline in the Elderly).

A logistic regression model was constructed taking physicians’ decisions as the dependent variables and adjusting for patient factors and physician impressions.

Results: The functional status reported by reliable informants and the mental status measured by validated instruments were not coincident with the physicians’ perception (functional status ?, 0.47; mental status ?, 0.26). A multivariate analysis showed that the age and the functional and mental status of patients, as perceived by the physicians, were the variables that better explained the physicians’ decisions. Conclusions: Physicians’ impressions on the functional and mental status of their patients significantly influenced their selection of patients for high-intensity treatments despite the fact that some of these impressions were not correct.

(C) 2010

Introduction

Advanced age has been found to be associated to increased mortality in critical care units [1,2]. However, it seems clear that such a negative outcome is not due to age itself but to a group of adverse factors, which often appear

* Corresponding author. Fax: +34 91 497 53 53.

E-mail address: [email protected] (A. Rodriguez-Molinero).

associated to age. Thus, in several studies, a poor premorbid functional status or the severity and type of the pathology that caused admission were found to be independent predictors of a negative outcome, whereas advanced age was not found to be an independent risk factor [3-5]. Therefore, some groups of elderly patients show good survival in the critical care units and maintain a good functional status and acceptable quality of life after discharge from hospital [6-9]. Conversely, other elderly patients do not seem to benefit from advanced life support so that such measures are futile [10,11].

0735-6757/$ – see front matter (C) 2010 doi:10.1016/j.ajem.2009.03.016

Unfortunately, no scale is currently available to select those elderly patients who will actually benefit from admission to critical care units. Actually, the most frequently used forecast indexes in these units do not consider the patients’ premorbid functional status and, consequently, seem to be inadequate for use in the Geriatric population [12]. Furthermore, the premorbid functional status of critical patients is usually not systematically recorded in the emergency services, where the initial decisions regarding their treatment are often made [13]. Thus, there is the possibility that patient selection for critical care treatments is not optimal, at least regarding functionality criteria.

Although elderly patients and their relatives should play a major role in decisions regarding the therapeutic effort, it has been shown that physicians are a key component in such decisions because they prescribe or dismiss aggressive interventions [14] Thus, a relationship has been observed between the act of limiting life support and the fact that the physician thinks that his or her patient has a less than 10% intensive care unit survival probability. This negative impres- sion predicted mortality even better than did organic dysfunc- tion, illness severity, or use of inotropes or vasopressors [15,16]. In view of the above considerations, it seems rather important to analyze the elements of the medical criterion on which physicians make their decisions for critical care treatment of their elderly patients. On the basis of the postulate that inaccurate information concerning elderly patients’ functional status may lead to suboptimal decisions, in this study, we analyzed the aggressive intervention plans of several emergency physicians. We have taken into account not only relevant elements such as comorbidity or patients’ premorbid functional status but also the physicians’ impression about their patients’ Baseline functional capacity, regardless of whether it was correct or incorrect. In this way, we attempted to identify the factors involved in physicians’ decision making regarding admission of elderly patients to critical care units; in addition, we intended to evaluate the appropriateness of such decisions in relation to the disability

degree of elderly patients.

The differences between physicians’, patients’, and relatives’ perceptions of the patients’ functional status have been previously identified in some studies based on transversal designs and small samples. However, to our knowledge, the influence that such differences have on physicians’ plans for intensive intervention on elderly patients at the ED setting has not been systematically analyzed [17,18].

Methods

Study participants and setting

This was a cross-sectional study conducted from July to November 2003 at 4 major Madrid university teaching hospitals, having a catchment area of 700 000 persons each.

This study included 101 elderly patients, their physicians, and their relatives. We included patients older than 80 years, and patients between 65 and 79 years, provided that the latter had at least 2 comorbid chronic conditions. Thus, elderly patients who were younger, more robust, and not having multiple conditions did not take part in the study. Participat- ing patients were selected based on chronic conditions that fulfilled the criteria described in the Dictionary of Health Services Management [19]. In particular, selected patients had the following conditions: cardiovascular conditions (congestive heart failure, rhythm abnormalities, central or peripheral vascular disease, hypertension), metabolic and endocrinologic disorders (obesity, hyperlipidemia, diabetes mellitus, thyroid diseases), tumors, chronic anemias, respira- tory conditions (chronic obstructive pulmonary disease, obstructive Sleep apnea, obliterans bronchiolitis), chronic kidney disease, bone and rheumatic diseases (gout, arthritis, osteoarthritis, osteoporosis), gastrointestinal diseases (chronic hepatitis, chronic gastritis, colonic diseases, gastroesophageal reflux disease), central nervous system degenerative disorders (Parkinson and related syndromes, dementias), and psychia- tric conditions (major Depressive disorder).

The sample was selected by numbering all patients admitted to emergencies on each day of the study and then selecting participants by means of a table of random numbers. The study days were chosen based on the pollster’s availability, and both weekdays and weekends were included. All elderly subjects who fulfilled the inclusion criteria were included, except for those who had no available informant, who failed to sign the informed consent, who had no clinical history of emergencies, or whose physician failed to meet the criteria outlined below.

Once a patient had been selected, one of the physicians declaring themselves responsible for the patient was required to participate. The highest ranking physician was selected, with those who had less than 1 year of experience or had already participated in the study in connection with another patient being excluded. All of the selected physicians except one consented to participate; consequently, another available physician who was of equal rank and also carried responsibility was therefore selected in his place. The participating physicians were blind to the objectives of the study. However, after taking the interview for the first time, physicians could easily infer the purposes of the study consequently altering their responses to subsequent inter- views. Therefore, every physician was only allowed to participate once, in relation to only one of his or her patients. Five purpose-trained medical interviewers were tasked with data collection. Even when the interviewers were physicians with experience in the emergency clinical practice, and were therefore very familiar with gathering information in the emergency services, they undertook a structured training program aimed at training them in correct data collection procedures. The program included theoretical and practical training on the methodology for managing the scales and measuring instruments to be used, as well as for

collecting secondary data from the patients’ emergency medical records. The data collected from the medical records were entered on a form designed by the researchers. The form included multiple-choice questions as well as questions with a blank space, where the answer could be freely expressed, depending on the kind of information to be gathered. All the interviewers were accompanied during their first day of fieldwork. In addition, they were provided with a consultation telephone number, where they could reach a supervisor researcher throughout the duration of the field- work. The supervisor researcher subsequently reviewed about 15% of the medical records. One hundred percent of the questionnaires were also reviewed to detect possible inconsistencies [20,21]. Only a few mistakes were detected, all of which could be successfully clarified by the interviewers. All the data were collected during the patients’ visit to the ED, once the initial medical assessment was performed, and the medical history was written.

According to the Spanish laws that regulate clinical research, given the observational nature of this study, explicit approval by an institutional review board was not required for conducting the present research.

Measures and interventions

First, each of the participating physicians was asked to decide whether the selected patient would be a candidate for intensive care treatment, in the event that he or she needed it (Fig. 1). We emphasize that the selected patients did not need intensive care treatment at the moment of the study; thus, the physicians’ plans were recorded as “advanced directives.” In particular, we asked the physicians to express their action plans by answering “yes” or “no” to the following questions included in a questionnaire: “If this patient had a cardior- espiratory arrest at this point in time, would you resuscitate him/her?” “If this patient required orotracheal intubation, would you propose him/her for admission to the MICU?” “If this patient had a coronary syndrome with ST elevation, would you propose him/her for Admission to the coronary care unit ? In making these decisions, physicians were not subject to a time limit and were free to consult Clinical histories, talk to patients, their families, or any other sources they deemed appropriate. Only after the physicians declared to be ready for the decision-making process were the decisions recorded and the questionnaires supplied.

The interviewers recorded the reasons that led the physicians to take their decisions, based on 2 questions administered via a questionnaire. One of the questions was open-ended (“What did you base your decision on?”) and the other was multiple-response, in which the physician could choose one or more of the following options: age, personal history, mental status, and functional status.

After having made their decisions on their patients, the physicians were asked about the impression that they had formed of their patients’ baseline functional capacity, through administration of 2 standard instruments, that is,

Fig. 1 Study data collection.

Katz index, for 6 Basic activities of daily living (BADL– bathing, dressing, toileting, transfers from bed to chair, continence and feeding), and the walking section of the Barthel index. These indexes are widely used in the clinical practice, as well as in research, for evaluation of elderly persons’ functional capacity. Both indexes classify elders into “dependent” or “independent” for every BADL according to detailed instructions provided by the authors [22,23]. If a physician was not sure about the patient’s degree of autonomy, the interviewer asked the physician to express his or her subjective impression about the matter. Likewise, physicians had the chance of stating their general impression regarding whether or not the patient had cognitive deterioration before visiting the emergency department (ED) (cognitive deterioration was defined as any alteration of Cognitive functions of sufficient magnitude to test positive on standard screening).

The interviewers then recorded all the patient’s past medical conditions from his or her clinical history, the current diagnosis or, in its absence, the reason for the medical visit. Subsequently, patients’ prior functional status was evaluated by administering Katz test for 6 BADL and the Barthel index question on walking to family members or informants who stated that they knew the patient. The functional capacity can be quantified by administering the Katz and Barthel indexes through a BADL observation procedure or by interviewing an informant [24,25]. Direct BADL observation has been generally considered to be the best method. However, when direct observation is not feasible, it is discussed whether the elderly patient or the relatives are more reliable as a source or information [25]. In this study, researchers considered that direct BADL observation during emergency situations would be neither feasible nor reliable to establish patient’s premorbid func- tional status (because this could be influenced by the presenting condition); on the same basis, they considered the information supplied by a proxy to be more reliable than

that supplied by the patient [17].

Lastly, patients’ prior mental status was assessed by administering the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE), validated for the Spanish population, to a family member or informant [26,27]. This tool has good internal validity in the general population (? = .95), acceptable test-retest validity (r = 0.75) [26], and good diagnostic correlation indexes with the Mini-mental state Examination [28], occasionally performing better than it, as indicated by Del-Ser et al [29]. It is considered to have 86% sensitivity, 92% specificity, 54% positive predictive value, and 91% negative predictive value. In every case, family members were asked to report the patient’s functional and cognitive status previous to the onset of the disease or decompensation that caused demand for emergency services. Patients’ current cognitive status was also evaluated by direct administration of Pfeiffer test validated for the Spanish-population Short Portable Mental Status Questionnaire (SPMSQ) [30]. The Test-retest reliability of this tool ranges between 0.82 and 0.85 [31], its sensitivity ranges between 68% and 82%, and its specificity ranges between 92% and 96% [32].

Statistical analysis

We carried out a descriptive analysis of the frequency of the different Therapeutic measures adopted by physicians, as well as the rationale behind such decisions, according to the same physicians. Patients’ functional status, as regarded by their physicians (Katz index administered to physicians), was compared to that reported by the relatives (Katz index administered to the relatives) by using the ? index for concordance.

Afterward, to ascertain which variables were explanatory of the diagnostic-therapeutic decisions targeted by this study, we constructed a logistic regression model taking physicians’ decisions vis-a-vis the different actions proposed (cardiopul-

monary resuscitation [CPR], admission to medical intensive care unit [MICU], admission to CCU) as the dependent variables. The model was adjusted for the following variables that were entered in different blocks: patient’s sociodemo- graphic profile (sex, age, educational level, marital status), patient comorbidity and history of cancer, patient’s cognition and patient’s functional status (IQCODE, SPMSQ, Barthel walking item, and Katz index reported by family members), diagnostic impression of cognitive deterioration, and diag- nostic impression of functional status on the part of the physician (Barthel walking item and Katz index estimated by physicians). Those cases in which the physicians declared to have based their opinion on moral issues or on the family’s or patient’s opinions were excluded from this analysis.

To evaluate the outcome that would have resulted from the physicians’ decisions, we conducted a descriptive analysis of the number and characteristics of patients with a good functional status (Katz 5-6) who were not considered candidates for intensive care and severely impaired patients (Katz 0-1) who were considered candidates for critical care. Finally, we analyzed the extent to which physicians underestimated the functional capacity of patients who would have not undergone intensive care (CPR, MICU, CCU) and the extent to which physicians overestimated the functional capacity of patients who would have received intensive care. To that end, we calculated the percentage of cases in which the functional capacity had been under- estimated and overestimated in the different situations and we analyzed the differences in such percentages between the

“admitted” and “rejected” groups.

All the statistical analyses were carried out with the SPSS

11.0 statistics package (SPSS, Chicago, Ill).

Results

The baseline characteristics of the study patients are summarized in Table 1. All data were obtained from valid informants, except for the patients’ cognitive situation at the moment of the study, which was evaluated by administering them the Pfeiffer test.

The mean age of the emergency physicians was 30.9 years (+-5.2 years) and the predominant sex was female (66.7%). Of the total, 42.6% specialized in internal medicine, 27.7% in family and community medicine, and the remaining in other medical specialties; 35.4% of physicians were hospital staff physicians and the remaining were resident physicians in their second to fifth years of training (the percentages of included consultant physicians in each hospital were 40%, 36.8%, 40%, and 20%).

Insofar as patients’ family members or informants were concerned, 66% were women and mean age was 56.1 years (+-12.6 years); 64.4% were children of patients, 18.8% were spouses, and 15.8% other family members; 1% of informants had no family relationship with the patients. In terms of

n (%)			n (%)
Sex		Katz
Male	60 (61.2)	Totally independent	34	(35.4)
Female	38 (38.8)	Totally dependent	8	(8.3)
Age (mean +- SD)	81.7 +- 7.3	Bathing
Educational level		Independent	55	(55.6)
No formal education	46 (47.4)	Dependent	44	(44.4)
Elementary education	47 (48.5)	Dressing
University education	4 (4.1)	Independent	75	(75.8)
Marital status		Dependent	24	(24.2)
Single	5 (5.1)	Toileting
Married	46 (46.9)	Independent	71	(71.7)
Widowed	46 (46.9)	Dependent	28	(28.3)
Separated/divorced	1 (1.0)	Transfers from bed to chair
Living		Independent	72	(73.5)
Alone	7 (7.2)	Dependent	26	(26.5)
In a residence	3 (3.1)	Continence
With spouse	35 (36.1)	Independent	44	(44.9)
With children	34 (35.1)	Dependent	54	(55.1)
With spouse and children	8 (8.2)	Feeding
Other	10 (10.3)	Independent	90	(91.8)
Pfeiffer test a		Dependent	8	(8.2)
Positive	43 (48.3)	Walking

schooling, 10.2% of family members had no formal education, 68.4% had elementary education, and 21.4% had university education.

Table 1 Sociodemographic, cognitive, and functional characteristics of the patient sample (n = 101)

Negative

46 (51,7)

Independent Dependent

59 (62.8)

35 (37.2)

^a Cognitive impairment screening test.

Of the total patients, physicians stated that they would perform CPR on 59.6%, admission to MICU in case intubation was needed on 48.5%, and propose 73.7% for admission to CCU.

The factors cited by physicians as being taken into account in their decisions regarding these patients were, in descending order of frequency, prior functional status (69.3%), age (41.6%), personal history (29.7%), current disease (17.8%), prior mental status (9.9%), subjective assessments of quality of life (4%), family or patient preferences (2%), social status (2%), and moral considera- tions (1% “this is a patient with the right to live”).

The concordance (?) between the respective Katz indices obtained from physicians and respondents was 0.47 (95% confidence interval, 0.38-0.57). This statistic, which indi- cates agreement between 2 observers, ranges from -1 to +1, with 1 representing complete agreement, 0 representing agreement by chance, and negative values representing complete disagreement [33]. In this case, we found only moderate agreement (the calculated power of the above comparison reached 92%) The concordance (?) between IQCODE obtained from the relatives and physicians’ perceptions of cognitive impairment was 0.26 (95% confidence interval, 0.06-0.45; power of the comparison, 95%), which represents low concordance.

Five cases were excluded from the multivariate analysis because the physicians claimed to have based their decisions on the patients’ or the relatives’ preferences, or on moral issues. The analysis was thus conducted on 74 cases with complete data.

Multivariate analysis revealed that the physicians’ impressions on the patients’ functional status (not the functional status as reported by an informant who cohabitates with the patient) was the only variable that could explain for the administration of CPR.

The patient’s mental status–as evaluated by the physician (not as measured by the administered test)–and the functional status as evaluated by the physician had a major influence on the decision of admitting patients to the MICU, whereas the patient’s age had a little influence. Similarly, admission to the CCU was mainly explained by the physicians’ impression on their patients’ functional status (Table 2).

When the physician’s impressions about the patient’s functional status and cognition were excluded from the analysis, only the patient’s age and history of neoplasia could explain, although very partially, some of the physicians’ decisions. Comorbidity, patient’s premobid functional status, and patient’s previous cognition (IQCODE, SPMSQ) did not explain any of the physicians’ decisions.

Forty patients with reliable informants were independent for bathing, dressing, toileting, transfers from bed to chair, feeding, and walking (continence was not taken into account). Twenty-four of these patients would have under- gone CPR or would have been admitted to the CCU or the

Blocks included	Independent variable	?	OR	Significance
CPR Block 1 + 2 + 3	Katz index estimated by physician	0.68	1.97 (1.16-3.35)	.012 a
Admission to MICU
Block 1	Age	-0.77	0.93 (0.86-1.00)	.056
Block 1 + 2	Age	-0.11	0.90 (0.82-0.99)	.024 a
Block 1 + 2 + 3	Age	-0.53	0.86 (0.75-0.98)	.025 a
	Cognition estimated by physician	2.73	15.38 (1.32-179.52)	.029 a
	Katz index estimated by physician	1.41	4.09 (1.81-9.25)	.001 a
Admission to CCU Block 1	Age	-0.11	0.90 (0.82-0.99)	.032 a
Block 1 + 2	Age	-0.12	0.89 (0.80-0.99)	.029 a
Block 1 + 2 + 3	Age	-2.77	0.76 (0.59-0.97)	.025 a
	Cognition estimated by physician	3.65	38.63 (0.89-1676.62)	.058
	Katz index estimated by physician	1.46	4.32 (1.70-10.95)	.002 a

MICU if it were necessary, whereas 16 patients would have been rejected for invasive measures. In the last subgroup, 5 patients had pathologies that could have worsen the prognosis, according to the physician’s criterion; for the remaining 11 patients, concordance between the functional status according to the physician’s opinion and according to the informant’s opinion was null (bisquare weighted ?, 0.08); thus, the physician underestimated the functional status of 8 of these 11 patients. The Wilcoxon test showed statistically significant differences between the Katz index estimated by the physician and that reported by the relatives (P b .05) Underestimation of the function as compared to the basal level was especially striking for the 4 patients who were not selected for any intensive care options (Table 3).

Table 2 Relevant results of multivariate analysis

Block 1 (patient’s Sociodemographic variables): sex, age, educational level, marital status. Block 2 (patient diseases and functional status): comorbidity, IQCODE, SPMSQ, Barthel walking item, Katz index. Block 3 (physician-related variables): cognition estimated by physician, Barthel walking item estimated by physician, Katz index estimated by physician.

^a Significant finding.

From the group of 17 patients with reliable informants who were dependent for five to six BADL, 12 (70.6%) patients would have undergone resuscitation or been

admitted to critical care units, this decision being attributable to the preference of the family in only one case. For 8 of the

11 remaining patients, the physician overestimated their functional status (Wilcoxon, P b .05; bisquare weighted ?, 0.034) (Table 4).

Of patients who would not have been resuscitated (CPR), 62.9% had better functional status than that estimated by their physicians; this percentage was much smaller (25.5%) among patients who would have been resuscitated (P b .001) (Table 5). Similar results were found after analysis of admissions to the MICU (64.4% cases of underestimated functional capacity among nonadmitted patients and 15.6% cases among admitted ones P b .001), and admissions to the CCU (70.8% underestimated functionality among rejected patients and 28.8% underestimated functionality among admitted ones; P b .001). Conversely, the candidates for intensive care proposed by the physicians systematically

Table 3 Baseline situation, real and estimated by physicians, in 11 independent patients not scheduled for intensive care

Table 4 Baseline situation, real and estimated by physicians, in 11 severely disabled patients scheduled for intensive care

Case no.	CPR	Admission to ICU	Admission to CCU	Katz index (physician)	Katz index (family)
1	No	No	No	1	5
2	No	No	No	1	6
3	No	No	No	0	6
4	No	No	No	0	6
5	Yes	No	No	2	6
6	No	No	Yes	6	5
7	Yes	No	Yes	4	6
8	Yes	No	Yes	5	6
9	Yes	No	Yes	6	6
10	No	Yes	Yes	2	5
11	No	Yes	Yes	4	6
				Mean, 2.82	Mean, 5.72
Katz index: 6 = Independent for all BADL; 0 = Dependent for all BADL.

Case no.	CPR	ICU	CCU	Katz index (physician)	Katz index (family)
1	Yes	Yes	Yes	6	1
2	Yes	Yes	Yes	4	0
3	Yes	Yes	Yes	4	1
4	Yes	Yes	Yes	4	1
5	Yes	Yes	Yes	4	0
6	Yes	No	No	1	1
7	No	No	Yes	2	0
8	No	No	Yes	0	1
9	No	No	Yes	0	1
10	No	No	Yes	2	1
11	No	No	Yes	1	0
				Mean, 2.54	Mean, 0.64
Katz index: 6 = independent for all BADL; 0 = dependent for all BADL.

Physician’s assessment of patient’s functional status	Patients rejected, % (n)	Patients admitted, % (n)	P
CPR
Overestimation	20.0 (7)	38.2 (21)	.069
Accurate	17.1 (6)	36.4 (20)	.05
Underestimation	62.9 (22)	25.5 (14)	b.001 a
MICU
Overestimation	13.3 (6)	48.9 (22)	b.001 a
Accurate	22.2 (10)	35.6 (16)	.163
Underestimation	64.4 (29)	15.6 (7)	b.001 a
CCU
Overestimation	4.2 (1)	40.9 (27)	b.001 a
Accurate	25.0 (6)	30.3 (20)	.624
Underestimation	70.8 (17)	28.8 (19)	b.001 a
a Significant finding.

corresponded to physician’s overestimation of their func- tional capacity (Table 5).

Table 5 Accuracy of physicians’ assessment of functional status in relation to their decisions for critical care treatment

Discussion

In this study, emergency physicians had to decide the therapeutic options for their patients in the event that they needed advanced life support or intensive care.

After having made their decisions, the physicians were requested to identify the factors that had influenced such decisions. Besides the factors identified by the physicians, the researchers involved in this study recorded objective information about a number of elements that might or should guide therapeutic decisions. Such a double approach allowed evaluating if the physicians’ decisions actually adjusted to the medical criteria that they identified as relevant. The analyzed factors included physician-related factors such as the physician’s impression about the patient’s functional status or cognition; this is a novel approach to an already scarcely studied issue.

In our study, physicians would have prescribed intensive care treatment in more than half the cases. The age and the premorbid functional status were the factors that physicians most frequently identified as relevant. The patient’s condition at the moment of the study and the pathologic background were also mentioned in a significant number of cases.

However, multivariate models did not confirm the influence of all the above factors on the physicians’ decisions (models included age, comorbidity, history of cancer and premorbid functional and mental status as reported by reliable informants). Surprisingly, in spite of the relevance of the premorbid functional status–as acknowledged by the physicians–this variable could not explain the physicians’ decisions independently; conversely, the influence of the patient’s age was confirmed. However, when the patient’s functional and mental status as evaluated by the physician

was included in the multivariate models, both variables did explain a significant proportion of the physicians’ decisions. Thus, coincident with their answers during the interview, a number of physicians did take the patients’ functional status into account for their decision making. However, the physicians’ perception of the functional status was signifi- cantly different from that of the relatives. Such a difference between the physicians’ and the informants’ criterion has been previously described [13,17], where the physicians’ perception was found to be less precise than that of the informant-relatives. In the present study, we used the information provided by relatives as the standard. All the included patients had an informant and those patients whose informants did not seem reliable to the interviewers were excluded from the analysis. Neither direct BADL observation (considered to be the best standard for comparison in other situations) nor administration of questionnaires to the elderly patients (more controversial) could be used to find out the premorbid functional status of patients at the ED because the Acute conditions that caused the patient to ask for emergency assistance could significantly influence the results [18]. However, using informants as the reference standard resulted in exclusion of elderly patients who were not accompanied by a reliable informant. Consequently, physicians’ actions in respect of patients who were unac- companied by an informant were not measured, but decisions would not be expected to be any more appropriate for subjects without than for subjects with an informant, particularly in the

case of patients with cognitive deterioration.

Physicians declared that the patient’s cognitive capacity was also a relevant factor influencing their decision making. Multivariate analysis confirmed that physicians’ impressions about the patients’ cognition were determinant in some decisions. However, much like that for the functional status, physicians’ recognition of cognitive decline was not accurate. Such inability of emergency physicians to recognize cognitive dysfunction or even simply act on such has been previously reported by other authors [34,35].

In view of our results, the relevance of the physician’s impression on the patient’s functional or mental status for decision making is evident, although such impression is not totally accurate [18]. Unfortunately, physicians’ partial knowl- edge of their patients’ functional status may lead to suboptimal selection of patients to be admitted to the critical care units. In this regard, a partial analysis of the most striking cases of our study is helpful. Both in the case of severely impaired patients who were considered candidates for intensive care and in independent patients who were not, the physician’s unfami- liarity with the patient’s functional status might have played a crucial role. A further hint in this direction is the fact that underestimation of the patient’s functional status was more frequent among patients who were not proposed for intensive care, whereas overestimation of the functional capacity was more frequent among patients who were selected for aggressive treatments. These findings highlight the possible influence of physician’s subjective considerations on the selection of patients for intensive treatments.

Restrictions in the access of elderly patients–as compared to persons in other age groups–to medical treatments or research studies have been previously reported [36-38]. Because our study included only elderly persons, comparison with emergency physicians’ decision making concerning younger persons is not possible. However, our results suggest that restrictions in the access of some elders to critical care units are mainly due to inaccurate judgment of their premorbid basal status and consequently therefore to erroneous patient classification. As a result, access to critical care units of patients with a priori better prognosis is often restricted, whereas access of patients with a priori worse prognosis is allowed.

Physicians’ decisions may be legitimately influenced by factors other than those considered here, such as the patient’s or the relatives’ preferences, the patient’s social situation, or the physician’s moral principles. However, it is surprising that the physicians rarely took these circumstances into account probably because, in this study, the acute complica- tions of the patients presented to the physicians were hypothetical and not real, something that could tempt physicians to take decisions based on clinical aspects in preference to other considerations.

One limitation of this study is its small sample size, which may pose problems of precision and generalizability. The sample size used is in part a consequence of the difficulty of enrolling 101 emergency physicians who had not previously participated in the study, because fulfilling the questionnaire could influence subsequent decisions concerning other patients.

Furthermore, a large proportion of the physicians participating in this study were undertaking specialization training. It can thus be speculated that they had not achieved the technical ability necessary for an adequate selection of patients to be admitted to the intensive care [39]. None- theless, we preferred to include these resident physicians– from their second year of training–because they support a large proportion of the emergency services burden and they must make independent decisions while on duty. Excluding such physicians would have biased the selection, resulting in a sample that was not representative of the reality of the selected hospitals’ (ascertainment bias) [40]. However, including resident physicians restricted the external validity of our study because our results cannot be extrapolated to hospitals without medical specialization training programs.

It did not prove feasible to use different interviewers to collect data from informants, physicians, and clinical histories, separately. The fact that the interviewers were not blinded to the responses of the different participants (physicians and informants) constituted a limitation because it could result in the occurrence of observer bias [41] (possible influence of the interviewers on the responses of the interviewed participants). We endeavored to minimize this limitation by imposing a striCT order on data collection, that is, only after collecting information from the physicians and the clinical histories could the questionnaires be

administered to the relatives, thereby preventing any influence on the data drawn from physicians and clinical histories. We also considered controlling for possible bias derived from the Hawthorne effect [42], namely, physicians could have improved their agreement with the relatives, or modified their intensive Care decisions, as a consequence of being observed in the frame of this study. However, because the physicians were blinded to the purposes of the study and to the relatives’ and patients’ answers about the patient’s functional and cognitive capacities (the Katz index and IQCODE reported by the relatives and the SPMSQ corresponding to the patient), no influence of the mentioned effect on the results was expected.

In this study, physicians forecast their plan of action with respect to complications that had not occurred. It is thus evident that, in the event of real complications, the final actions might well be different from these intended plans. We did not measure final actions; yet, we do not expect that these would necessarily be better than those that were forecast, seeing as the latter were drawn up under less stressful conditions and with more time available. This is the reason why it is recommended that DNR orders be recorded in clinical histories [14]. The researchers were not concerned with measuring the severity of the conditions affecting the patients included in this study because the complications proposed to the physicians, being hypothetical, were not related to the severity of the actual conditions. However, there is the possibility that the severity of the present diseases affecting patients influenced the physicians’ decisions, a possibility which was not controlled for.

Elderly persons between 65 and 79 years of age were not included in this study unless they had 2 or more chronic pathologies. Thus, younger elders and healthy elders were not represented in our sample, which poses limitations to the generalizability of our results. This is due to the authors’ interest in focusing the study on the frailest elderly persons for whom medical decisions are more complex and interesting.

Overall, our results suggest that physicians do acknowl- edge the relevance of elderly patients’ premorbid functional and cognitive status for their decision making. As a matter of fact, their decisions are clearly influenced by their knowl- edge of patients’ functional and cognitive status. However, because their knowledge of these aspects is often inaccurate, the intensive care planned by emergency physicians for their elderly patients is not in all cases consistent with the latter’s functional status. Of all the variables studied, physicians’ perceptions of their patients’ functional status was the factor that best explained the planning of care by physicians.

Acknowledgments

We would like to thank Concepcion Jimenez Rojas for kindly supplying initial bibliography of interest, and Ainhoa

Esteve Arrien, Cristina Horrillo Garcia, and Sebastian Martin Balbueno for their invaluable help in collecting study data.

References

1. Cohen IL, Lambrinos J. Investigating the impact of age on outcome of mechanical ventilation using a population of 41,848 patients from a statewide database. Chest 1995;107(6):1673-80.
2. Dewar DM, Kurek CJ, Lambrinos J, et al. Patterns in costs and outcomes for patients with prolonged mechanical ventilation under- going tracheostomy: an analysis of discharges under diagnosis-related group 483 in New York State from 1992 to 1996. Crit Care Med 1999;27(12):2640-7.
3. Bo M, Massaia M, Raspo S, et al. Predictive factors of in-hospital mortality in older patients admitted to a medical intensive care unit. J Am Geriatr Soc 2003;51(4):529-33.
4. Rockwood K, Noseworthy TW, Gibney RT, et al. One-year outcome of elderly and young patients admitted to intensive care units. Crit Care Med 1993;21(5):687-91.
5. Somme D, Maillet JM, Gisselbrecht M, et al. Critically ill old and the oldest-old patients in intensive care: short-and long-term outcomes. Intensive Care Med 2003;29(12):2137-43.
6. Konopad E, Noseworthy TW, Johnston R, et al. quality of life measures before and one year after admission to an intensive care unit. Crit Care Med 1995;23(10):1653-9.
7. Montuclard L, Garrouste-Orgeas M, Timsit JF, et al. Outcome, functional autonomy, and quality of life of elderly patients with a long- term Intensive care unit stay. Crit Care Med 2000;28(10):3389-95.
8. Nierman DM, Schechter CB, Cannon LM, et al. outcome prediction model for very elderly critically ill patients. Crit Care Med 2001;29 (10):1853-9.
9. Van Den NN, Vogelaers D, Afschrift M, et al. Intensive care for very elderly patients: outcome and risk factors for in-hospital mortality. Age Ageing 1999;28(3):253-6.
10. Awoke S, Mouton CP, Parrott M. Outcomes of skilled cardiopulmon- ary resuscitation in a long-term-care facility: futile therapy? J Am Geriatr Soc 1992;40(6):593-5.
11. Bilsky GS, Banja JD. Outcomes following cardiopulmonary resuscita- tion in an acute rehabilitation hospital. Clinical and ethical implica- tions. Am J Phys Med Rehabil 1992;71(4):232-5.
12. de Rooij SE, bu-Hanna A, Levi M, et al. Factors that predict outcome of intensive care treatment in very elderly patients: a review. Crit Care 2005;9(4):R307-14.
13. Salvi F, Morichi V, Grilli A, et al. The elderly in the emergency department: a critical review of problems and solutions. Intern Emerg Med 2007;2(4):292-301.
14. Goodlin SJ, Zhong Z, Lynn J, et al. Factors associated with use of cardiopulmonary resuscitation in seriously ill hospitalized adults. JAMA 1999;282(24):2333-9.
15. Cook D, Rocker G, Marshall J, et al. Withdrawal of mechanical ventilation in anticipation of death in the intensive care unit. N Engl J Med 2003;349(12):1123-32.
16. Rocker G, Cook D, Sjokvist P, et al. Clinician predictions of intensive care unit mortality. Crit Care Med 2004;32(5):1149-54.
17. Elam JT, Graney MJ, Beaver T, et al. Comparison of subjective ratings of function with observed functional ability of frail older persons. Am J Public Health 1991;81(9):1127-30.
18. Gerson LW, Blanda M, Dhingra P, et al. Do elder emergency department patients and their informants agree about the elder’s functioning? Acad Emerg Med 2001;8(7):721-4.
19. Owings Mills M. Dictionary of health services management. National Health Pub. ISBN: 0932500560; 1987.
20. Gilbert EH, Lowenstein SR, Koziol-McLain J, et al. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996;27(3):305-8.
21. Reisch LM, Fosse JS, Beverly K, et al. Training, quality assurance, and assessment of medical record abstraction in a multisite study. Am J Epidemiol 2003;157(6):546-51.
22. Katz S, Ford AB, Moskowitz RW, et al. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963;185:914-9.
23. Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. Md State Med J 1965;14:61-5.
24. Chen MH, Hsieh CL, Mao HF, et al. Differences between patient and proxy reports in the assessment of disability after stroke. Clin Rehabil 2007;21(4):351-6.
25. Weinberger M, Samsa GP, Schmader K, et al. Comparing proxy and Patients’ perceptions of patients’ functional status: results from an outpatient geriatric clinic. J Am Geriatr Soc 1992;40(6):585-8.
26. Jorm AF, Jacomb PA. The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): socio-demographic correlates, reliability, validity and some norms. Psychol Med 1989;19(4):1015-22.
27. Morales Gonzalez JM, Gonzalez-Montalvo JI, Del Ser QT, et al. Validation of the S-IQCODE: the Spanish version of the informant questionnaire on cognitive decline in the elderly. Arch Neurobiol (Madr ) 1992;55(6):262-6.
28. Jorm AF, Scott R, Cullen JS, et al. Performance of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) as a screening test for dementia. Psychol Med 1991;21(3):785-90.
29. Del-Ser T, Morales JM, Barquero MS, et al. Application of a Spanish version of the “Informant Questionnaire on Cognitive Decline in the Elderly” in the clinical assessment of dementia. Alzheimer Dis Assoc Disord 1997;11(1):3-8.
30. Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc 1975;23(10):433-41.
31. Lesher EL, Whelihan WM. Reliability of mental status instruments administered to nursing home residents. J Consult Clin Psychol 1986;54(5):726-7.
32. Del Ser T, Pena Casanova J. Evaluacion neuropsicologica y funcional de la demencia. Barcelona: JR Prous Editores; 1994.
33. Sim J, Wright CC. The kappa statistic in reliability studies: use, interpretation, and sample size requirements. Phys Ther 2005;85 (3):257-68.
34. Hustey FM, Meldon SW. The prevalence and documentation of impaired mental status in elderly emergency department patients. Ann Emerg Med 2002;39(3):248-53.
35. Hustey FM, Meldon SW, Smith MD, et al. The effect of mental status screening on the care of elderly emergency department patients. Ann Emerg Med 2003;41(5):678-84.
36. Dohrenwend PB, Fiesseler FW, Cochrane DG, et al. Very young and elderly patients are less likely to receive narcotic prescriptions for clavicle fractures. Am J Emerg Med 2007;25 (6):651-3.
37. Giugliano RP, Camargo Jr CA, Lloyd-Jones DM, et al. Elderly patients receive less aggressive medical and invasive management of unstable angina: potential impact of practice guidelines. Arch Intern Med 1998;158(10):1113-20.
38. Lee PY, Alexander KP, Hammill BG, et al. Representation of elderly persons and women in published randomized trials of acute coronary syndromes. JAMA 2001;286(6):708-13.
39. McNamara RM, Rousseau E, Sanders AB. Geriatric EMergency medicine: a survey of practicing emergency physicians. Ann Emerg Med 1992;21(7):796-801.
40. Delgado-Rodriguez M, Llorca J. Bias. J Epidemiol Community Health

2004;58(8):635-41.

1. Johannes CB, Crawford SL, McKinlay JB. Interviewer effects in a cohort study. Results from the Massachusetts Women’s Health Study. Am J Epidemiol 1997;146(5):429-38.
2. Wickstrom G, Bendix T. The “Hawthorne effect”–what did the original Hawthorne studies actually show? Scand J Work Environ Health 2000;26(4):363-7.