OCR Text |
Show physicians1 Tables 2 and 3 display urban and rural/frontier physicians' self-ratings of the influence of specific factors on the decision to admit patients in the CHF, pneumonia, and cellulitis scenarios. When considering medical reasons alone, physicians were more likely to admit in the CHF and pneumonia scenarios than in the cellulitis scenario. When individual non-medical factors were introduced, variability in responses was noted. Average ratings of the impact of these non-medical factors were significantly different from "neither more nor less likely to admit" (p<0.0001 by Hotelling's T-squared) for the sets of both pro-outpatient and pro-inpatient factors. Therefore, we examined more closely which non-medical factors were influential in which diagnostic scenarios. Three non-medical factors appeared relatively uninfluential to the decision to hospitalize. Over 85% of physicians reported that having a consultant available by telemedicine would not alter the decision to admit a patient in any diagnostic scenario. 61-75% of physicians felt that the presence or absence of a local specialist would not alter the decision, depending on scenario. Similarly, over half of physicians felt that more or less comprehensive insurance beyond Medicare would not change the admissions decision in any scenario. These factors were dropped from the pro-outpatient and pro-inpatient summary scales. Naming the scales "pro-outpatient" and "pro-inpatient" was validated by two analyses in addition to factor analysis. Both the pro-outpatient summary scale (excellent family support, close to care, and family prefers outpatient care) and the pro-inpatient summary scale (poor family support, far from care, and family prefers to hospitalize) had means significantly different from the null "no more or less likely to admit" in the expected direction (p=0.0001 by ANOVA.) Also, physicians were more likely to admit by self-report for pro-inpatient vs. pro-outpatient factors in each diagnostic scenario (p=0.0001). Average propensity to hospitalize, based on the pro-inpatient summary scale, did not vary significantly across diagnoses (p=0.873 by repeated measures MANOVA.) However, the pro-outpatient summary scale displayed significant differences for different diagnoses 1 Three Utah Counties had no resident physician during the conduct of this study. (p=0.0001 by repeated measures MANOVA.) Likelihood of admission based on pro-outpatient non-medical factors was reduced most for the cellulitis case. The average urban physician's score on the pro-outpatient ACSC summary scale was 2.6 compared to a rural average of 2.5 (p=0.564 by ANOVA.) Rural physicians were slightly more likely to hospitalize these ACSC cases when pro-inpatient non-medical factors came into play (mean=4.0) than were urban physicians (mean=3.8; p=0.041 by ANOVA.) Discussion This study found that physicians' self-reported propensity to hospitalize elderly patients with three ambulatory care sensitive conditions varied with specific non-medical factors. Rural physicians were slightly more likely than urban physicians to increase their propensity to hospitalize based on pro-inpatient non-medical factors such as lack of family support at home and distance from care. It is of note that respondents did not answer uniformly "no" when asked whether non-medical factors would have made the physician more or less likely to admit a patient in a scenario to the hospital, although there was high consensus on individual factors. Recognizing the limitations of response based on self-report (Campos-Outcalt, 1993), physicians clearly and consistently differentiated non-medical factors in their responses. Further, physicians in different geographic locales responded in significantly different ways depending on which non-medical factors and which diagnoses were involved. For example, a majority of physicians reported "no influence" of more or less comprehensive insurance beyond Medicare, availability of a specialist, or availability of consultation by telemedicine. Lack of interest in telemedicine may reflect its limited reach in Utah at this time, low applicability to these disease processes, attitudes of distrust or insecurity regarding its use, or otherwise insufficient provider readiness on one side of the consultation or the other. Similarly, comprehensiveness of insurance was not identified by this study as a pro-outpatient or pro-inpatient factor, and this may reflect a limitation of self-report. Urban and rural/frontier physicians did not differ on ratings of whether patients in each of the CHF, pneumonia, and cellulitis scenarios should be admitted, when only medical factors were involved. However, only the cellulitis case achieved the ambivalence desired for this baseline scenario. This may have limited the 22 |