Reported Home Health Agency Referrals by Internists and Family Phvsicians J
J
Peter A. Boling, MD,*Joseph M. Keenan, MD,t Joanne Schwartzberg, MD,$ Sheldon M. Retchin, MD, MSPH,* Lorayn Olson, PhD,§ and Mindy Schneiderman, PhD§ Objective: To evaluate the frequency of home health agency referrals (HHRs) by internists and family physicians. Design: Telephone survey of a randomly selected, nationally representative,stratified physician sample. Participantsand Setting: One thousand one hundred sixtyone interviews with 576 family physicians and 585 internists selected from the American Medical Association Physician Masterfile. Main Results: Most respondents (88%) reported making HHRs (mean for those making HHRs = 43/year). Physicians with 2 48 annual HHRs (n = 315) reported a mean of 2.6 hours/week in home care telephone management and 2.1 hours/week on related paperwork. Rural internistsand family physicians (n = 230) reported less availability of several types of non-physician home health services than non-rural respondents (n = 931), yet rural physicians were more likely to refer patients to home health agencies. Using multivariate
linear regression, the reported frequency of HHRs was significantly related to rural practice location, number of homebound patients, proportion of geriatric patients, number of house calls, graduation from a U.S. or Canadian medical school, physician knowledge of community resources, and physician experience either as a medical director, a member of the board of directors, or a consultant for a home health agency. Conclusions: Internists and family physicians who work at least 10 hours per week in ambulatory care report making approximately three home health agency referrals per month and spending substantialamounts of time coordinatinghome health agency care. Despite reportingless availability of many home health agency services, rural physicians report greater involvement than non-rural physicians in the delivery of home care. J Am Geriatr Soc 401241-1249,1992
ccess to physician care and availability of professional home care services are vital to the care of homebound persons. This is particularly true in rural areas where the physician density is low, home health agencies are fewer, distances traveled to receive care are greater, and transportation may be less available.' There are now over 1million chronically homebound persons in the United States.' The steadily increasing number of homebound Americans has been associated with a dramatic increase in the volume of home care services and greater technical sophistication in home care.3e These home care services are multi-faceted, including some typically delivered by home health agencies such as nursing care, physical therapy, occupational and speech therapies, and medical social work. Home care also includes other services, such as those delivered by personal care agencies and durable medical equipment companies. In 1987, 5.9 million or 2.5 percent of the US population received some home care service^.^ New needs for home health agency care develop in 3.2 percent of the elderly within a 2-year period.6 Between 1969 and 1984, Medicare home health agency visits rose from 8.5 million to 40.3 million.2
Recognizing the needs of the homebound, the American Medical Ass~ciation,~ the American College of Physicians,' and the American Geriatrics Society' have issued position statements favoring increased use of home health care and urging physicians to take more active roles. Physician oversight is required for most non-physician home health services. Yet, while physicians rely heavily on home health agencies as the mainstay of medical home care, many physicians lack capability in this area." Also, despite the expansion of home care, there has been no nationally representative study of physician interactions with home health agencies or physician time spent in supervision of home health agency care. While many specialties are involved in home care, we limited this study to the two largest physician groups involved in adult primary care to permit a statistically valid comparison by specialty. Therefore, we studied physicians' self-reported behaviors and attitudes by surveying a national sample of internists and family physicians with active office practices.
A
From the "Division of Geriatric Medicine, Department of Medicine, Virginia Commonwealth University/Medical College of Virginia, Richmond, Virginia; TDepartment of Family Medicine, University of Minnesota, Minneapolis, Minnesota; and $Department of Geriatric Health and §Department of Survey Research, American Medical Association, Chicago, Illinois. These data were presented, in part, at the meeting of the Gerontologic Society of America, November 1991, San Francisco, CA. Research supported, in part, by a grant from Caremark, Inc. and by the Department of Geriatric Health at the American Medical Association. The views in this paper are those of the authors and no offiaal endorsement by the American Medical Association is intended nor should be inferred. Address correspondence to Peter A. Bohg, MD, Box 102 MCV Station, Richmond, VA 23298-0102. ]AGS 40:1241-1249, 1992 0 1992 by the American Geriatrics Society
METHODS The study population was 35,572 family physicians and 73,652 internists, including general internists and subspecialists, in the American Medical Association (AMA) Physician Masterfile. The Masterfile is the most comprehensive national physician database in the United States. It contains demographic data, educational background, and current practice information for both AMA members and non-members. Excluded from the study were physicians over age 75, osteopathic physicians, physicians who were not officebased, and those employed by the Federal government. Random samples of 1,100 physicians each were se0002-8614/92/$3.50
1242
BOLING ET AL
lected from the specialties of internal medicine and family practice, thus over-sampling family physicians to support an intended comparison by specialty. A detailed telephone interview was developed based, in part, on two previous physician home care surveys.", l 2 Interviews were conducted with the physicians by the AMA Department of Survey Research between June and November of 1990. Physicians gave actual numeric estimates for some practice characteristics, such as the number of homebound patients they followed, their annual number of house calls and/or home health agency referrals, and the percent of older patients in their practices. Attitude questions used a four-point Likert-type scale. An average interview took 15 minutes. The definition of homebound was: "an individual who can leave home only with considerable or taxing effort, or with the assistance of another person or device." Four physicians who were deceased and 62 retired physicians were considered ineligible. Another 132 physicians, who reported less than 10 hours per week in outpatient care or that their specialty was neither internal medicine nor family medicine, were also considered ineligible. Subtracting these 198 physicians leaves a sample of 2,002 physicians. Twenty physicians were out of the country during the study period, and 197 physicians could not be located after a careful search. Without these 217 unlocatable physicians, the remaining sample includes 1,785 physicians. Nineteen percent of these 1,785 physicians were located but did not respond to the telephone calls, and another 16 percent of the 1,785 physicians refused to participate. Interviews were completed by 1,161 of these 1,785 physicians. The completion rate is 65 percent if the 21 7 unlocatable physicians are deemed ineligible, or 58 percent if they are deemed non-respondents. Data on the number of hours worked by physicians in various capacities were added from the AMA's Record of Physicians' Professional Activities, a rotating census of all U.S. physicians, in which one-third of the physicians are surveyed each year. To evaluate representatives, respondents' characteristics were compared with those of the total study population, ie, all Masterfile physicians meeting study criteria. There were no differences in sex, age, size of county in which the practice was located, or present employment (eg, solo vs group practice). There were also no significant differences between respondents and non-respondents in age, sex, practice location (rural vs urban), present employment, country of medical school graduation, or board certification status. Threecomponent weights were calculated to adjust the data for three respondent characteristics which were not representatively distributed, comparing respondents to the Masterfile. Respondents included a greater proportion of AMA members, a smaller proportion of physicians from the Northeast region, and, because of intentional oversampling, a larger proportion of family physicians. Factors associated with the number of annual home health agency referrals by physicians were then analyzed. First, respondents were partitioned into three
IAGS-DECEMBER 1992-VOL. 40, NO.12
referral frequency groups according to the annual number of home health agency referrals: "low" (0-1 1 referrals), "intermediate" (12-47 referrals), and "high" (248 referrals). Those physicians reporting zero referrals did not differ in their responses from those reporting between one and eleven referrals, and these,, therefore, were analyzed as a single group. Categorical variables were analyzed using the chi-square statistic, comparing the three physician groups. Second, linear regression was used to compare continuous independent variables across these three physician groups. (Linear regression was used here because the ANOVA procedure in SAS software does not permit weights). Weighted data were used for all comparisons described above. Finally, an unweighted general linear model was used to analyze the relative effects of multiple independent variables on home health agency referrals. When missing values were encountered for variables used in the multivariate regression model, the missing data were replaced by imputed values, using median values for continuous variables and mode values for Likert-type attitude variables. Because of its policy relevance, availability of home health agency services in rural areas was also analyzed using the chi-square statistic. After finding similar results using several different population density thresholds, rural location was defined to include physicians practicing in a county not considered part of an SMSA or a potential SMSA and having a population of less than 50,000. Where appropriate, Likert variables and continuous variables were collapsed to create binary variables to facilitate data analysis and presentation.
RESULTS In all, 1024 respondents (88 percent) reported one or more annual home health agency referrals (mean = 43 referrals/year among those who reported making referrals). These respondents also reported spending 2.1 hours per week on telephone supervision of home care and 1.5 hours per week on home care paperwork. There were 429 physicians in the "low" referral group, 417 physicians in the "intermediate" frequency group, and 315 physicians in the "high" referral frequency group. Shown in Table 1 are physicians' personal and professional characteristics according to annual home health agency referrals. Referral frequency was not related to physician age (mean = 45.4 f 10 years), physician specialty, number of hours employed each week by a medical school (mean = 2.5 f 11 hours), or the type of available home health agencies. Also, referral frequency was not associated with physician preference for a specific home health agency type. Referral frequency was associated with the type of practice, but there was no clear pattern in this association. Increased referral frequency was significantly associated with graduation from a US or Canadian medical school, rural location, and service by the physician as medical director, consultant, or member of the board of directors for a home health agency. The relationship of physicians' clinical practice char-
JAGS-DECEMBER 1992-VOL. 40, NO.12
HOME HEALTH AGENCY REFERRALS
1243
TABLE 1. HOME HEALTH AGENCY REFERRALS ACCORDING TO PERSONAL AND PROFESSIONAL CHARACTERISTICS OF PHYSICIANS****# Annual Home Care Referral Frequency
Characteristics Male sex Graduation from other than a US or Canadian medical school Physician specialty Family medicine Internal medicine Practice type Solo practice Group practice Other Rural practice location Has been an HHA medical director, on the HHA board of directors, or has been an HHA consultant Availability of HHA‘s by agency type For profit only Not-for-profit only Both types
Low
Intermediate
High
(0-11 Referrals) ( n = 429) (%)
(12-47 Referrals) ( n = 417) (%)
(>47 Referrals) (n = 315) (%)
345 (80) 113 (26)
374 (90) 62 (15)
280 (89)** 51 (16)**
208 (48) 223 (52)
181 (43) 235 (57)
138 (44) 176 (56)
147 (43) 121 (35) 75 (22) 68 (16) 24 ( 8 )
123 (34) 174 (49) 60 (17) 86 (21) 59 (14)
110 (39) 121 (43)* 52 (18) 76 (24)* 67 (21)*
44 (19) 22 (10) 161 (71)
60 (17) 54 (16) 229 (67)
55 (21) 31 (12) 179 (68)
67 (17) 97 (24) 238 (59)
35 (13) 76 (25) 191 (63)
38 (14) Hospital-based only 66 (25) Community-based only 166 (61) Both types * P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.001 comparing characteristic across the three referral frequency groups. *** Colunm totals for some variables are altered by rounding after weighting. # Percents shoroii in parentheses are column percents.
acteristics with home health agency referral frequency is shown in Table 2 . Referral frequency was strongly associated with the number of work hours, the number of patient care hours, the proportion of older patients in the practice, whether or not the physician makes house calls, the number of homebound patients, and
time spent on home health care paperwork or telephone work. The availability and use of several categories of home health agency services are displayed in Figure 1 according to frequency of home health agency referrals. Referral frequency was significantly associated with
TABLE 2. HOME HEALTH AGENCY REFERRAL FREQUENCY ACCORDING TO PHYSICIANS CLINICAL PRACTICE CHARACTERISTICS Annual Home Care Referral Frequency
Characteristics Office visit charge in dollars Total work hours per week Patient care hours per week Percent of physician’s patients who are 65 and older Number of homebound patients in physician’s practice Number of house calls by physician in the past year Number of home health agency referrals in the past year Hours per week on telephone managing home health agency care Hours spent on forms for home health care per week Makes house calls (yes/no)
Low
Intermediate
High
(0-11 Referrals) (n = 429)
(12-47 Referrals) ( n = 417)
(>47 Referrals) (n = 315)
(SD) (SD) (SD) (SD)
38 60 52 36
(16) (17) (16) (25)
(SD) 12 (38)
(13) (14) (13) (21)
35 65 57 47
(lo)** (16)*** (15)***
19 (36)
36
(50)***
37 62 55 43
(20)***
(SD)
7
(40)
8
(19)
(SD)
3
(4)
24
(9)
(SD)
1.4 (2)
2.1 (3)
2.6 (2),**
(SD)
1.2 (2)
1.3 (1)
2.1 (2)***
(%) 144
(34)
* P C 0.05 comparing characteristic across the three referral frequency groups.
** P C 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.001 comparing characteristic across the three referra[ frequency groups.
236
(57)
12 (33)* 95
199
(59)***
(63)***
1244
BOLING ET AL
IAGS-DECEMBER 1992-VOL. 40, NO. 12
AVAILABILITY HOMEMAKER NURSING CARE
**
13T PT. SPEECY
SYCIAL WORK tz3LIANCES A3GRATORY
-=!.IF
HOSPICE
I
I
0
20
40
60
80
100
.PERCENT RESPONDING "EASILY AVAILABLE"
U S hOMEMAKER**+ hU2SING CARE+** CT. 77. SPEECH
**
X C I A L WORK
***
F??LIANCES
***
-ABORATORY
***
I
HOME HOSPICE *** 20
0
40
60
80
100
PERCENT RESPONDING "ALWAYS OR OFTEN U S E
. .. .*.
p < .05
< 12 REFERRALS/YR
p c .01
1 2 . 4 6 REFERRALS /YR
p c ,001
> 46 REFERRALS/YR
FIGURE 1. Home Health Agency Service.
easier availability of nursing services and with availability of homemaker services. There was no significant association between home care referral frequency and the availability of other home health services, but there
were consistent associations between referral frequency and reported use of all service types. Physicians were also questioned regarding availability of and preferences for community-based home health agencies, rather than hospital-based agencies, and for-profit rather than not-for-profit agencies. Of 538 physicians who had both profit and non-profit types available, 183 physicians (34%) preferred notfor-profit agencies, 84 physicians (16%) preferred forprofit agencies, and 271 physicians (50%) expressed no preference. Of 579 physicians who had both community-based and hospital-based agencies available, 128 physicians (22%) preferred community-based agencies, 206 physicians (36%) preferred hospitalbased agencies, and 245 physicians (42%) expressed no preference. Shown in Table 3 are agreement rates from 1,056 physician respondents to 15 possible clinical and social indications for home care services, according to the frequency of home health agency referrals. Respondents rated the likelihood of making a home health agency referral for a given indication. The number of respondents who agreed or strongly agreed with each indication are presented in the table. For most indications, physicians making more home health agency referrals more often agreed that a given indication would prompt a referral. For nine indications these differences were statistically significant. Four attitudes showed no significant association with the frequency of home health agency referrals: personal satisfaction from the referral (number in agreement with the indication for referral = 37 percent), management of bedridden pregnancy complications (25 percent agreed), patient had been in physician's care for a long time (75 percent agreed), and pressure from the family (69 percent agreed). General attitudes toward home care are presented in
TABLE 3. HOME HEALTH AGENCY REFERRAL FREQUENCY AND NUMBER OF PHYSICIANS WHO AGREE OR STRONGLY AGREE WITH INDICATIONS FOR HOME HEALTH AGENCY SERVICES# Annual Home Care Referral Frequency
Indications
Low (0-11 Referrals) ( n = 324) (%I
195 (74) Patient could not afford van or ambulance 218 (79) Transport available but too difficult for patient 221 (79) Allow patient to stay home (vs. nursing home) Improve compliance with medical 205 (72) care plan 134 (51) Post-surgical care 190 (67) Need to assess home or family situation 114 (41) Need to assess acute medical problems 209 (74) Provide terminal care 193 (68) Manage chronic medical problems * P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.005 comparing characteristic across the three referral frequency groups. #Percents shown in parentheses are column percents.
Intermediate (12-47 Referrals) ( n = 417) (%)
High (>47 Referrals) (n = 315) (%)
280 (69)
227 (78)*
323 (79)
260 (86)*
364 (89)
269 (88)**
326 (79)
265 (85)**
247 (63) 320 (77)
173 (62)** 272 (88)***
215 (52)
184 (59),,*
348 (84) 335 (81)
271 (86)*** 241 (78)***
IAGS-DECEMBER 1992-VOL. 40, NO. 12
HOME HEALTH AGENCY REFERRALS
Table 4. The frequency of referrals was associated with only four attitudes. Compared with physicians making more referrals, those making fewer referrals agreed slightly more often that families could learn to perform care at home. Larger differences were seen in physician beliefs that they can provide adequate care at home and that malpractice risks are not increased by home care; agreement with both attitudes increased significantly with referral frequency. Finally, physicians' agreement that they have sufficient knowledge of community resources to plan home health care showed a strong, significant association with increased referral frequency. Respondents were presented with a hypothetical homebound patient calling with an "acute but not emergent" problem and were asked to choose one strategy to address the problem. These results are shown in Table 5. Transporting patients to the office or the emergency room was slightly more likely among physicians reporting a low frequency of referrals. Physicians with higher referral frequencies less often chose "other" strategies such as telephone management and were much more likely to make a home health agency referral to resolve the situation. Use of a house call by the physician or an office-based nurse practitioner was equally common among the three physician groups. The differences in choice of strategies across the three referral frequency groups were significant. Because of the importance of rural location, home
1245
health agency service availability was analyzed by practice location. Compared with physicians in nonrural locations ( n = 931), physicians in rural locations (n = 230) reported similar availability of nursing services (84 percent vs 85 percent respectively, NS). However, compared with non-rural settings, occupational, physical, and speech therapy were significantly less often considered easily available in rural settings (52 percent vs 68 percent; P < 0.001), as were medical social services (51 percent vs 68 percent; P < 0.001), and appliance or equipment services (73 percent vs 83 percent; P < 0.001). Rural physicians were also more likely than non-rural physicians to report frequent use of homemaker services (64 percent vs 54 percent, P < 0.01) and laboratory services (84 percent vs 7 7 percent, P < 0.05). The results of the linear regression model are shown in Table 6. Rural location, greater physician knowledge of community resources, graduation from a US or Canadian medical school, a higher proportion of patients over age 64, a greater number of homebound patients, experience as a medical director, member of the board of directors, or consultant for a home health agency, and a greater number of house calls were all independently associated with a greater number of home health agency referrals. The physician's weekly hours spent in patient care, physician sex, specialty, age, employment by a medical school, the perceived malpractice risk in home care, and the perceived avail-
TABLE 4. HOME HEALTH AGENCY REFERRAL FREQUENCY AND NUMBER OF PHYSICIANS WHO AGREE OR STRONGLY AGREE WITH ATTITUDES TOWARD HOME CARE# Annual Home Care Referral Frequency
Physician Attitude
Low (0-11 Referrals) (n = 429) (Or,)
I can provide adequate medical care in 189 (47) the home Most patients/families can learn to per391 (94) form care at home Most families can learn to perform com176 (43) plex care at home Malpractice risk in home care is no 164 (44) greater than in hospital/clinic I have sufficient knowledge of commu244 (59) nity services to personally plan/deliver home health care Home environment/family support is 256 (66) usually adequate for care at home All home care agencies offer same qual117 (32) ity/range of services I am satisfied with service quality of 319 (87) home care providers in my area Physicians should use home care agencies 330 (87) more Non-MD home care services adequately 109 (41) covered by third party payors With readily available home nursing (RN, 230 (57) aide), most MD house calls unnecessary 269 (67) I am too busy with office/hospital practice to make house calls P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.005 comparing characteristic across the three referral frequency groups. # Percents shown in parentheses are column percents.
Intermediate (12-47 Referrals) (n = 417) (%I
High (>47 Referrals) (n = 315) (%)
206 (51)
181 (60)*
373 (90)
278 (89)**
185 (45)
145 (46)
216 (56)
155 (53)*
309 (74)
263 (85)***
288 (73)
225 (74)
129 (33)
84 (29)
366 (88)
284 (89)
324 (84)
240 (83)
116 (41)
103 (46)
238 (59)
183 (60)
249 (60)
197 (63)
1246
BOLING ET AL
IAGS-DECEMBER 1992-VOL. 40, NO. 12
TABLE 5. HOME HEALTH AGENCY REFERRAL FREQUENCY ACCORDING TO PHYSICIANS' STRATEGY FOR CARE OF ACUTE PROBLEMS IN HOMEBOUND PATIENTS** Annual Home Care Referral Frequency Low (0-11Referrals) Physician's Strategy
(n = 429) (%)
Intermediate (12-47 Referrals) (n = 417) (%)
High (>47 Referrals) (n = 315) (%)
Given a homebound patient with an acute, non-emergent problem would you:* Send patient to hospital ER by ambu73 (18) 56 (14) 43 (14) lance Transport patient to office by ambu83 (20) 76 (18) 52 (17) lanceIvan Schedule physician house call 53 (13) 49 (12) 32 (10) Send office nurse, NP, or PA on 9 (2) 17 (4) 11 (4) house call Schedule a home care nursing visit by 56 (14) 75 (18) 102 (33) an outside agency Schedule a home care nursing visit 11 (3) 12 (3) 3 (1) and MD house call Other (eg, phone management) 122 (30) 126 (31) 70 (22) *- P < 0.001 cornparing tl7e proportion of physicians wi70 select a given strategy across the three referral frequent!/ groups by chi-square analysis
(7
x 3 tabld. '"I;
Pcrcerifagcs iii parentheses are coluniii percents.
TABLE 6. LINEAR REGRESSION MODEL OF HOME HEALTH AGENCY REFERRAL FREQUENCY WITH PHYSICIAN AND PRACTICE CHARACTERISTICS# Standardized Variable Beta Coefficient Intercept Greater hours per week employed by medical school Physician belief that malpractice risk is not increased in home care vs other setting Physician specialty, internist Greater physician age Physician sex, female Greater physician hours of patient care per week Home health agency nursing service is easily available Physician has served as home health agency director, board member, or consultant Physician comfort with own knowledge of community resources Physician graduation from North American medical school Rural practice location Higher percentage of geriatric patients in practice Greater number of house calls per year Greater number of homebound patients
0.000 -0.002 0.003 0.016 0.012 0.018 0.025 0.045 0.067* 0.078* 0.082* 0.096** 0.1 63***
0.133*** 0.156***
R ' = 0.126. * P < 0.05 probability of statistically significant independent contribution to model. *:* P < 0.01 probability of statistically significant independent contribution to model. **:l P < 0.001 probabilit!y of statistically significant independent contribution to niodel. # Negative beta co-efficient indicates a negative effect of the predictor variable on home health agency referral frequency.
ability of nursing services did not independently predict referrals. Using more extreme definitions of "rural" status in the model did not alter these findings.
DISCUSSION According to the 1987 National Medical Expenditure S ~ r v e y 5.9 , ~ million Americans used some home care services, altogether costing $12.1 billion. The Brooking Institution has projected that total home health care expenditures, measured in constant dollars, will triple in the next three decades.I3 Since physician oversight
is required for most home care services, physician involvement has fiscal relevance as well as clinical importance for the community-based health care team. In this national survey of family physicians and internists, most respondents reported some involvement with home health agency care. While 37 percent of respondents made less than 12 annual home care referrals, the typical physician reported making about three referrals per month, and 27 percent of respondents made frequent referrals, averaging 95 per year. These data reflect both the extent of the community
JAGS-DECEMBER 1992-VOL. 40, NO. 12
need and the reliance of physicians on home health agencies. Considered in fiscal terms, given that one typical episode of home health agency care involves 25 visits, at a cost of $66 per visit,I4 the average physician would annually order almost $60,000 worth of home health agency services. This study helps define some determinants of physician home care referral practice. Physicians’ personal and professional characteristics are often related to their practice patterns. For example, physician house call fre uency has been associated with physician specialty’ and with physician age.” However, in this survey, home care referral frequency was not related to most personal or professional characteristics of the physicians. Home care referrals showed a stronger association with practice profiles. The higher referral frequency for physicians with more work hours may reflect busier clinical practices. The larger proportion of geriatric patients and number of homebound patients among those making more referrals was also expected. Likewise, it was not surprising to find that those 168 (14 percent) respondents who had worked as a home health agency medical director, consultant, or member of a home health agency board of directors made more referrals. Often physicians selected for this role have extensive home care experience. Practical constraints precluded evaluating physicians’ vested interests as a motive. The finding, confirmed by the regression model, that home care referrals were less common among physicians graduating from medical schools outside of North America was unexpected. This might reflect unfamiliarity with community resources or culturally determined practice patterns. However, there are many possible interpretations, and further investigation would be required to further explore this issue. Rural physicians reported more home care referrals, and rural location was a strong independent predictor in the regression model. While service unavailability and logistic problems might tend to limit rural referrals, other practice demands and distances between homes might increase referrals, substituting for direct physician service in rural locations. Although nursing care was equally available to rural and non-rural physicians, other services such as physical therapy, social work, and appliances (eg, wheelchairs) were far less often considered readily available in rural areas. Some of these services are provided by part-time contract professionals who may be less available to the nonprofit agencies, which are often the only available agencies in rural areas. Overall, lack of service availability may be a significant problem in rural settings where home care appears to be a particularly important strategy for care delivery. Most of the physicians endorsed a wide variety of indications for home health agency referrals. However, compared with those who made fewer home care referrals, physicians making frequent referrals were much more likely to use home health agencies in many situations. Thus physicians who frequently use home health agencies may be more comfortable with an expanded health care delivery role for home health
HOME HEALTH AGENCY REFERRALS
1247
agencies. There were large differences in referral fre-. quency related to attitudes reflecting chronic care, such as the need to assess the home, improving compliance, providing terminal care, and management of chronic medical problems. Moreover, home health agencies were more often used in response to urgent care situations by physicians reporting more frequent referrals. Thus the tendency of physicians who make more frequent referrals to use home health agencies more actively appears to span the spectrum of acute and chronic care. Although almost 90 percent of respondents were satisfied with the quality of the available services, less than one-third of respondents considered the quality of home health agency services to be uniform. However, despite the variation in perceived service quality, referral frequency was not related to perceived service quality. There was no association of referral frequency with the type of agency available (eg, for-profit or hospital-based). Nor was referral frequency related to attitudes reflecting the physicians’ freedom from other responsibilities to make house calls or the belief that home health agency services may substitute for physician house calls. This finding is consonant with the observation that referral frequency increased with house call frequency, suggesting that home health agency care is not simply being substituted for physician care of the homebound. Concern about inadequate family support might also affect physician use of home health agencies. Although less than half of the physicians felt that most families can learn complex caregiving tasks, two-thirds considered the home environment adequate for home care, and referral frequency did not vary with these attitudes. Overall, more than 90 percent of physicians agreed that most families can learn to perform care at home. Three attitudes suggest a relationship between physician knowledge and home care referral practice. These are: the physicians’ belief that they can provide adequate medical care in the home; physician comfort with their knowledge of community resources; and belief that there is no increase in malpractice risk when ordering home care. The importance of physicians’ selfperceived knowledge of community resources was confirmed by the regression model. Whether physician home care knowledge is a result of active home care practice, or the practice a consequence of education, cannot be resolved by these data. The data did not support the hypothesis that availability of home health agency services drives referrals. Although nursing services and homemaker services were more often easily available to physicians making more referrals, the differences were modest. Most service categories were considered available by most respondents. Conversely, 15 percent of respondents did not consider nursing services readily available, a finding of potential concern considering the extent of the community need. Not surprisingly, use of the various individual home health services (eg, physical therapy or social work) was consistently greater among physicians making
1248
IAGS-DECEMBER 1992-VOL. 40, NO. 12
BOLING ET AL
more referrals. Use of the full spectrum of home health agency services may reflect recognition of the needs of the homebound by home health agency personnel. Often, home health agency personnel develop the home care plan of treatment, because many physicians are insufficiently familiar with home care regulations or patients’ home care needs to write treatment plans.’* The physician’s role in overseeing home health agency care deserves special comment. Third-party regulations require written physician approval for most home care services, and home health agency professionals often rely on physicians for telephone consultation during the course of treatment. Although time spent on case management did not increase linearly with home health agency referral frequency in this study, there was a consistent increase in time spent as referral frequency increased. Survey respondents making large numbers of referrals reported spending nearly 5 of 65 hours worked per week in activity related to the supervision of home care. The average physician spent 2.1 hours per week on telephone supervision of home care, and another 1.5 hours per week on related paperwork. This problem is compounded by new Medicare regulations which increased the requirement for direct physician involvement in completing forms authorizing use of durable medical equipment.I6 Despite the recent creation of charge codes specific to home care case management, Medicare intermediaries refuse to pay case management charges. Thus, while consuming substantial time, most physician work in support of home health agency care is un-reimbursed, undermining future physician motivation. The linear regression model places the major findings of this survey in perspective. Availability of nursing services, concern about malpractice, and physician work hours did not independently predict home health referral frequency. Moreover, physician specialty did not predict referral frequency, despite the fact that family medicine residency curriculum guidelines place greater emphasis on home care.17 On the other hand, variables reflecting the number of homebound patients in the physicians‘ practices were the strongest predictors of referrals, suggesting that patient needs play a substantial role in driving home health referrals by physicians. The importance of home care in rural practice is also confirmed by controlling for availability of nursing services and other factors. Finally, physicians’ self-perceived knowledge of community resources independently predicts home care referrals, suggesting an important educational agenda. Some limitations of this study should be noted. Although the response rate is comparable to that in most physician surveys,’8,l9 the possibility of response bias must be recognized. We could not evaluate attitudinal differences, but we did evaluate demographic variables and adjusted for the few differences between the respondents and the study population. Also, inherent in any survey reporting attitudes is the possibility that attitudes evolve to match behavior and that changes in attitudes will not necessarily predict changed behaviors. However, the respondents were clearly selective in their endorsement of attitudes.
Moreover, the attitudes are consonant with those previously reported in smaller samples.”, Third, we cannot verify the reported physician home health agency referral frequency and related time involvement. Finally, concern about response rate and cost forced a number of difficult choices during the survey design. The goal was to study both home visits and home care referral practice is broadly as possible. Thus, we could not insert many more detailed questions about such policy-relevant but complicated issues as the intermittency of the Medicare skilled home care benefit, differences between uses of homemakers and personal care aides, different levels of family caregiver sophistication, or physicians’ home care experience during training. The findings are also limited to family physicians and internists. We cannot comment on the home health agency referral practice of osteopaths, pediatricians, surgeons, psychiatrists, general practitioners, or obstetrician-gynecologists. However, family physicians and internists provide most of adult primary care, so we believe the study describes the majority of adult home care referral practice. Due to the length of the survey, data regarding house calls and “high tech” home care, such as home intravenous therapy, are being reported separately. Featured predominantly in the explosive growth of home care are professional services provided by home health agencies under physician supervision. This survey shows that most internists and family physicians make a large number of home health agency referrals and spend an appreciable amount of uncompensated time in work related to this referral practice. This finding is of particular importance, since relative work is now being measured and valued while reimbursement is increasingly being bundled. A substantial minority of physicians have less active home care referral practices, which may relate to several factors, such as lack of knowledge about home care or lack of homebound patients. Since home care is a key part of our society’s response to the needs of its rapidly increasing homebound population, these findings indicate a need for policy reform and educational initiatives to encourage continuation and enhancement of the physician‘s role in care of the homebound.
REFERENCES 1. Rowland D, Lyons
B. Triple jeopardy: Rural, poor, and uninsured. HSR
1989;23:975-1004. 2. Stone RI, Murtaugh CM. The elderly population with chronic functional
disability: Implications for home care eligibility. Gerontologist 1990;30: 491-496. 3. Committee on Ways and Means, U.S.House of Representatives. In: The 1991 Green Book. Washington, DC: U.S. Government Printing Office, 1991, p 150. 4. Spiegel AD. The growth of homecare. In: Spiegel AD. Home Health Care, 2nd Ed. Owings Mills, M D National Health Publishing, 1987, pp 329366. 5. Short PF, Leon J. Use of home and community services by persons ages 65 and older with functional difficulties. (DHSS Publication No. (PHS) 90-3466). National Medical Expenditure Survey Research Findings 5, Agency for Health Care Policy and Research. Rockville, MD. Public Health Service. 6. Branch LG, Wetle TT, Scherr PA et al. A prospective study of incident comprehensive medical home care use among the elderly. Am j Public Health 1988;78:255-259. 7. Council on Scientific Affairs of the American Medical Association. Home care in the 1990s. JAMA 1990;263:1241-1244.
IAGS-DECEMBER 1992-VOL. 40, NO. 12 8. American College of Physicians. Home health care. Ann Intern Med 1986; 105:484-490. 9. Public Policy Committee of the American Geriatrics Society. Home care and home care reimbursement. J Am Geriatr SOC1989;37:1065-1066. 10. Keenan JM, Hepburn KW, Ripsin CM et al. Home care agencies perceptions of Minnesota physician home care practice and skills: Results of a statewide survey. Fam Med 1992;24:142-144. 11. Boling PA, Retchin SM, Ellis J et al. The influence of physician specialty on housecalls. Arch Intern Med 1990;150:2333-2337. 12. Keenan JM, Bland CJ, Webster L, Myers S. The home care practice and attitudes of Minnesota family physicians. J Am Geriatr Soc 1991;39: 1100-1 104. 13. Rivlin A, Wiener J. Caring for the Disabled Elderly, Who Will Pay? Washington, D.C.: The Brookings Institution, 1988, p 42.
HOME HEALTH AGENCY REFERRALS
1249
14. National Association for Home Care. Basic Statistics About Home Care1991. Washington, DC, p 2. 15. Schueler MS, Hams DL, Goodenough GK et al. House calls in Utah. West J Med 1987;147:92-94. 16. Medicare Carriers Manual. Health Care Finance Administration Publication no. 6, Sections 60-64. Revised August, 1989. 17. Directory of Graduate Medical Education Programs. Section 2: Requirements for accreditation of programs. Chicago, IL: American Medical Association, 1989, pp 26-27. 18. Gunn WJ, Rhodes IM. Physician response rates to a telephone survey: Effects of monetary incentive on response level. Pub1 Opinion Q 1981; 45:109-115. 19. Babbie ER. Survey Research Methods. Belmont, CA: Wadsworth Publishing CO. Inc., 1973, p 165.
J
Peter A. Boling, MD,*Joseph M. Keenan, MD,t Joanne Schwartzberg, MD,$ Sheldon M. Retchin, MD, MSPH,* Lorayn Olson, PhD,§ and Mindy Schneiderman, PhD§ Objective: To evaluate the frequency of home health agency referrals (HHRs) by internists and family physicians. Design: Telephone survey of a randomly selected, nationally representative,stratified physician sample. Participantsand Setting: One thousand one hundred sixtyone interviews with 576 family physicians and 585 internists selected from the American Medical Association Physician Masterfile. Main Results: Most respondents (88%) reported making HHRs (mean for those making HHRs = 43/year). Physicians with 2 48 annual HHRs (n = 315) reported a mean of 2.6 hours/week in home care telephone management and 2.1 hours/week on related paperwork. Rural internistsand family physicians (n = 230) reported less availability of several types of non-physician home health services than non-rural respondents (n = 931), yet rural physicians were more likely to refer patients to home health agencies. Using multivariate
linear regression, the reported frequency of HHRs was significantly related to rural practice location, number of homebound patients, proportion of geriatric patients, number of house calls, graduation from a U.S. or Canadian medical school, physician knowledge of community resources, and physician experience either as a medical director, a member of the board of directors, or a consultant for a home health agency. Conclusions: Internists and family physicians who work at least 10 hours per week in ambulatory care report making approximately three home health agency referrals per month and spending substantialamounts of time coordinatinghome health agency care. Despite reportingless availability of many home health agency services, rural physicians report greater involvement than non-rural physicians in the delivery of home care. J Am Geriatr Soc 401241-1249,1992
ccess to physician care and availability of professional home care services are vital to the care of homebound persons. This is particularly true in rural areas where the physician density is low, home health agencies are fewer, distances traveled to receive care are greater, and transportation may be less available.' There are now over 1million chronically homebound persons in the United States.' The steadily increasing number of homebound Americans has been associated with a dramatic increase in the volume of home care services and greater technical sophistication in home care.3e These home care services are multi-faceted, including some typically delivered by home health agencies such as nursing care, physical therapy, occupational and speech therapies, and medical social work. Home care also includes other services, such as those delivered by personal care agencies and durable medical equipment companies. In 1987, 5.9 million or 2.5 percent of the US population received some home care service^.^ New needs for home health agency care develop in 3.2 percent of the elderly within a 2-year period.6 Between 1969 and 1984, Medicare home health agency visits rose from 8.5 million to 40.3 million.2
Recognizing the needs of the homebound, the American Medical Ass~ciation,~ the American College of Physicians,' and the American Geriatrics Society' have issued position statements favoring increased use of home health care and urging physicians to take more active roles. Physician oversight is required for most non-physician home health services. Yet, while physicians rely heavily on home health agencies as the mainstay of medical home care, many physicians lack capability in this area." Also, despite the expansion of home care, there has been no nationally representative study of physician interactions with home health agencies or physician time spent in supervision of home health agency care. While many specialties are involved in home care, we limited this study to the two largest physician groups involved in adult primary care to permit a statistically valid comparison by specialty. Therefore, we studied physicians' self-reported behaviors and attitudes by surveying a national sample of internists and family physicians with active office practices.
A
From the "Division of Geriatric Medicine, Department of Medicine, Virginia Commonwealth University/Medical College of Virginia, Richmond, Virginia; TDepartment of Family Medicine, University of Minnesota, Minneapolis, Minnesota; and $Department of Geriatric Health and §Department of Survey Research, American Medical Association, Chicago, Illinois. These data were presented, in part, at the meeting of the Gerontologic Society of America, November 1991, San Francisco, CA. Research supported, in part, by a grant from Caremark, Inc. and by the Department of Geriatric Health at the American Medical Association. The views in this paper are those of the authors and no offiaal endorsement by the American Medical Association is intended nor should be inferred. Address correspondence to Peter A. Bohg, MD, Box 102 MCV Station, Richmond, VA 23298-0102. ]AGS 40:1241-1249, 1992 0 1992 by the American Geriatrics Society
METHODS The study population was 35,572 family physicians and 73,652 internists, including general internists and subspecialists, in the American Medical Association (AMA) Physician Masterfile. The Masterfile is the most comprehensive national physician database in the United States. It contains demographic data, educational background, and current practice information for both AMA members and non-members. Excluded from the study were physicians over age 75, osteopathic physicians, physicians who were not officebased, and those employed by the Federal government. Random samples of 1,100 physicians each were se0002-8614/92/$3.50
1242
BOLING ET AL
lected from the specialties of internal medicine and family practice, thus over-sampling family physicians to support an intended comparison by specialty. A detailed telephone interview was developed based, in part, on two previous physician home care surveys.", l 2 Interviews were conducted with the physicians by the AMA Department of Survey Research between June and November of 1990. Physicians gave actual numeric estimates for some practice characteristics, such as the number of homebound patients they followed, their annual number of house calls and/or home health agency referrals, and the percent of older patients in their practices. Attitude questions used a four-point Likert-type scale. An average interview took 15 minutes. The definition of homebound was: "an individual who can leave home only with considerable or taxing effort, or with the assistance of another person or device." Four physicians who were deceased and 62 retired physicians were considered ineligible. Another 132 physicians, who reported less than 10 hours per week in outpatient care or that their specialty was neither internal medicine nor family medicine, were also considered ineligible. Subtracting these 198 physicians leaves a sample of 2,002 physicians. Twenty physicians were out of the country during the study period, and 197 physicians could not be located after a careful search. Without these 217 unlocatable physicians, the remaining sample includes 1,785 physicians. Nineteen percent of these 1,785 physicians were located but did not respond to the telephone calls, and another 16 percent of the 1,785 physicians refused to participate. Interviews were completed by 1,161 of these 1,785 physicians. The completion rate is 65 percent if the 21 7 unlocatable physicians are deemed ineligible, or 58 percent if they are deemed non-respondents. Data on the number of hours worked by physicians in various capacities were added from the AMA's Record of Physicians' Professional Activities, a rotating census of all U.S. physicians, in which one-third of the physicians are surveyed each year. To evaluate representatives, respondents' characteristics were compared with those of the total study population, ie, all Masterfile physicians meeting study criteria. There were no differences in sex, age, size of county in which the practice was located, or present employment (eg, solo vs group practice). There were also no significant differences between respondents and non-respondents in age, sex, practice location (rural vs urban), present employment, country of medical school graduation, or board certification status. Threecomponent weights were calculated to adjust the data for three respondent characteristics which were not representatively distributed, comparing respondents to the Masterfile. Respondents included a greater proportion of AMA members, a smaller proportion of physicians from the Northeast region, and, because of intentional oversampling, a larger proportion of family physicians. Factors associated with the number of annual home health agency referrals by physicians were then analyzed. First, respondents were partitioned into three
IAGS-DECEMBER 1992-VOL. 40, NO.12
referral frequency groups according to the annual number of home health agency referrals: "low" (0-1 1 referrals), "intermediate" (12-47 referrals), and "high" (248 referrals). Those physicians reporting zero referrals did not differ in their responses from those reporting between one and eleven referrals, and these,, therefore, were analyzed as a single group. Categorical variables were analyzed using the chi-square statistic, comparing the three physician groups. Second, linear regression was used to compare continuous independent variables across these three physician groups. (Linear regression was used here because the ANOVA procedure in SAS software does not permit weights). Weighted data were used for all comparisons described above. Finally, an unweighted general linear model was used to analyze the relative effects of multiple independent variables on home health agency referrals. When missing values were encountered for variables used in the multivariate regression model, the missing data were replaced by imputed values, using median values for continuous variables and mode values for Likert-type attitude variables. Because of its policy relevance, availability of home health agency services in rural areas was also analyzed using the chi-square statistic. After finding similar results using several different population density thresholds, rural location was defined to include physicians practicing in a county not considered part of an SMSA or a potential SMSA and having a population of less than 50,000. Where appropriate, Likert variables and continuous variables were collapsed to create binary variables to facilitate data analysis and presentation.
RESULTS In all, 1024 respondents (88 percent) reported one or more annual home health agency referrals (mean = 43 referrals/year among those who reported making referrals). These respondents also reported spending 2.1 hours per week on telephone supervision of home care and 1.5 hours per week on home care paperwork. There were 429 physicians in the "low" referral group, 417 physicians in the "intermediate" frequency group, and 315 physicians in the "high" referral frequency group. Shown in Table 1 are physicians' personal and professional characteristics according to annual home health agency referrals. Referral frequency was not related to physician age (mean = 45.4 f 10 years), physician specialty, number of hours employed each week by a medical school (mean = 2.5 f 11 hours), or the type of available home health agencies. Also, referral frequency was not associated with physician preference for a specific home health agency type. Referral frequency was associated with the type of practice, but there was no clear pattern in this association. Increased referral frequency was significantly associated with graduation from a US or Canadian medical school, rural location, and service by the physician as medical director, consultant, or member of the board of directors for a home health agency. The relationship of physicians' clinical practice char-
JAGS-DECEMBER 1992-VOL. 40, NO.12
HOME HEALTH AGENCY REFERRALS
1243
TABLE 1. HOME HEALTH AGENCY REFERRALS ACCORDING TO PERSONAL AND PROFESSIONAL CHARACTERISTICS OF PHYSICIANS****# Annual Home Care Referral Frequency
Characteristics Male sex Graduation from other than a US or Canadian medical school Physician specialty Family medicine Internal medicine Practice type Solo practice Group practice Other Rural practice location Has been an HHA medical director, on the HHA board of directors, or has been an HHA consultant Availability of HHA‘s by agency type For profit only Not-for-profit only Both types
Low
Intermediate
High
(0-11 Referrals) ( n = 429) (%)
(12-47 Referrals) ( n = 417) (%)
(>47 Referrals) (n = 315) (%)
345 (80) 113 (26)
374 (90) 62 (15)
280 (89)** 51 (16)**
208 (48) 223 (52)
181 (43) 235 (57)
138 (44) 176 (56)
147 (43) 121 (35) 75 (22) 68 (16) 24 ( 8 )
123 (34) 174 (49) 60 (17) 86 (21) 59 (14)
110 (39) 121 (43)* 52 (18) 76 (24)* 67 (21)*
44 (19) 22 (10) 161 (71)
60 (17) 54 (16) 229 (67)
55 (21) 31 (12) 179 (68)
67 (17) 97 (24) 238 (59)
35 (13) 76 (25) 191 (63)
38 (14) Hospital-based only 66 (25) Community-based only 166 (61) Both types * P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.001 comparing characteristic across the three referral frequency groups. *** Colunm totals for some variables are altered by rounding after weighting. # Percents shoroii in parentheses are column percents.
acteristics with home health agency referral frequency is shown in Table 2 . Referral frequency was strongly associated with the number of work hours, the number of patient care hours, the proportion of older patients in the practice, whether or not the physician makes house calls, the number of homebound patients, and
time spent on home health care paperwork or telephone work. The availability and use of several categories of home health agency services are displayed in Figure 1 according to frequency of home health agency referrals. Referral frequency was significantly associated with
TABLE 2. HOME HEALTH AGENCY REFERRAL FREQUENCY ACCORDING TO PHYSICIANS CLINICAL PRACTICE CHARACTERISTICS Annual Home Care Referral Frequency
Characteristics Office visit charge in dollars Total work hours per week Patient care hours per week Percent of physician’s patients who are 65 and older Number of homebound patients in physician’s practice Number of house calls by physician in the past year Number of home health agency referrals in the past year Hours per week on telephone managing home health agency care Hours spent on forms for home health care per week Makes house calls (yes/no)
Low
Intermediate
High
(0-11 Referrals) (n = 429)
(12-47 Referrals) ( n = 417)
(>47 Referrals) (n = 315)
(SD) (SD) (SD) (SD)
38 60 52 36
(16) (17) (16) (25)
(SD) 12 (38)
(13) (14) (13) (21)
35 65 57 47
(lo)** (16)*** (15)***
19 (36)
36
(50)***
37 62 55 43
(20)***
(SD)
7
(40)
8
(19)
(SD)
3
(4)
24
(9)
(SD)
1.4 (2)
2.1 (3)
2.6 (2),**
(SD)
1.2 (2)
1.3 (1)
2.1 (2)***
(%) 144
(34)
* P C 0.05 comparing characteristic across the three referral frequency groups.
** P C 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.001 comparing characteristic across the three referra[ frequency groups.
236
(57)
12 (33)* 95
199
(59)***
(63)***
1244
BOLING ET AL
IAGS-DECEMBER 1992-VOL. 40, NO. 12
AVAILABILITY HOMEMAKER NURSING CARE
**
13T PT. SPEECY
SYCIAL WORK tz3LIANCES A3GRATORY
-=!.IF
HOSPICE
I
I
0
20
40
60
80
100
.PERCENT RESPONDING "EASILY AVAILABLE"
U S hOMEMAKER**+ hU2SING CARE+** CT. 77. SPEECH
**
X C I A L WORK
***
F??LIANCES
***
-ABORATORY
***
I
HOME HOSPICE *** 20
0
40
60
80
100
PERCENT RESPONDING "ALWAYS OR OFTEN U S E
. .. .*.
p < .05
< 12 REFERRALS/YR
p c .01
1 2 . 4 6 REFERRALS /YR
p c ,001
> 46 REFERRALS/YR
FIGURE 1. Home Health Agency Service.
easier availability of nursing services and with availability of homemaker services. There was no significant association between home care referral frequency and the availability of other home health services, but there
were consistent associations between referral frequency and reported use of all service types. Physicians were also questioned regarding availability of and preferences for community-based home health agencies, rather than hospital-based agencies, and for-profit rather than not-for-profit agencies. Of 538 physicians who had both profit and non-profit types available, 183 physicians (34%) preferred notfor-profit agencies, 84 physicians (16%) preferred forprofit agencies, and 271 physicians (50%) expressed no preference. Of 579 physicians who had both community-based and hospital-based agencies available, 128 physicians (22%) preferred community-based agencies, 206 physicians (36%) preferred hospitalbased agencies, and 245 physicians (42%) expressed no preference. Shown in Table 3 are agreement rates from 1,056 physician respondents to 15 possible clinical and social indications for home care services, according to the frequency of home health agency referrals. Respondents rated the likelihood of making a home health agency referral for a given indication. The number of respondents who agreed or strongly agreed with each indication are presented in the table. For most indications, physicians making more home health agency referrals more often agreed that a given indication would prompt a referral. For nine indications these differences were statistically significant. Four attitudes showed no significant association with the frequency of home health agency referrals: personal satisfaction from the referral (number in agreement with the indication for referral = 37 percent), management of bedridden pregnancy complications (25 percent agreed), patient had been in physician's care for a long time (75 percent agreed), and pressure from the family (69 percent agreed). General attitudes toward home care are presented in
TABLE 3. HOME HEALTH AGENCY REFERRAL FREQUENCY AND NUMBER OF PHYSICIANS WHO AGREE OR STRONGLY AGREE WITH INDICATIONS FOR HOME HEALTH AGENCY SERVICES# Annual Home Care Referral Frequency
Indications
Low (0-11 Referrals) ( n = 324) (%I
195 (74) Patient could not afford van or ambulance 218 (79) Transport available but too difficult for patient 221 (79) Allow patient to stay home (vs. nursing home) Improve compliance with medical 205 (72) care plan 134 (51) Post-surgical care 190 (67) Need to assess home or family situation 114 (41) Need to assess acute medical problems 209 (74) Provide terminal care 193 (68) Manage chronic medical problems * P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.005 comparing characteristic across the three referral frequency groups. #Percents shown in parentheses are column percents.
Intermediate (12-47 Referrals) ( n = 417) (%)
High (>47 Referrals) (n = 315) (%)
280 (69)
227 (78)*
323 (79)
260 (86)*
364 (89)
269 (88)**
326 (79)
265 (85)**
247 (63) 320 (77)
173 (62)** 272 (88)***
215 (52)
184 (59),,*
348 (84) 335 (81)
271 (86)*** 241 (78)***
IAGS-DECEMBER 1992-VOL. 40, NO. 12
HOME HEALTH AGENCY REFERRALS
Table 4. The frequency of referrals was associated with only four attitudes. Compared with physicians making more referrals, those making fewer referrals agreed slightly more often that families could learn to perform care at home. Larger differences were seen in physician beliefs that they can provide adequate care at home and that malpractice risks are not increased by home care; agreement with both attitudes increased significantly with referral frequency. Finally, physicians' agreement that they have sufficient knowledge of community resources to plan home health care showed a strong, significant association with increased referral frequency. Respondents were presented with a hypothetical homebound patient calling with an "acute but not emergent" problem and were asked to choose one strategy to address the problem. These results are shown in Table 5. Transporting patients to the office or the emergency room was slightly more likely among physicians reporting a low frequency of referrals. Physicians with higher referral frequencies less often chose "other" strategies such as telephone management and were much more likely to make a home health agency referral to resolve the situation. Use of a house call by the physician or an office-based nurse practitioner was equally common among the three physician groups. The differences in choice of strategies across the three referral frequency groups were significant. Because of the importance of rural location, home
1245
health agency service availability was analyzed by practice location. Compared with physicians in nonrural locations ( n = 931), physicians in rural locations (n = 230) reported similar availability of nursing services (84 percent vs 85 percent respectively, NS). However, compared with non-rural settings, occupational, physical, and speech therapy were significantly less often considered easily available in rural settings (52 percent vs 68 percent; P < 0.001), as were medical social services (51 percent vs 68 percent; P < 0.001), and appliance or equipment services (73 percent vs 83 percent; P < 0.001). Rural physicians were also more likely than non-rural physicians to report frequent use of homemaker services (64 percent vs 54 percent, P < 0.01) and laboratory services (84 percent vs 7 7 percent, P < 0.05). The results of the linear regression model are shown in Table 6. Rural location, greater physician knowledge of community resources, graduation from a US or Canadian medical school, a higher proportion of patients over age 64, a greater number of homebound patients, experience as a medical director, member of the board of directors, or consultant for a home health agency, and a greater number of house calls were all independently associated with a greater number of home health agency referrals. The physician's weekly hours spent in patient care, physician sex, specialty, age, employment by a medical school, the perceived malpractice risk in home care, and the perceived avail-
TABLE 4. HOME HEALTH AGENCY REFERRAL FREQUENCY AND NUMBER OF PHYSICIANS WHO AGREE OR STRONGLY AGREE WITH ATTITUDES TOWARD HOME CARE# Annual Home Care Referral Frequency
Physician Attitude
Low (0-11 Referrals) (n = 429) (Or,)
I can provide adequate medical care in 189 (47) the home Most patients/families can learn to per391 (94) form care at home Most families can learn to perform com176 (43) plex care at home Malpractice risk in home care is no 164 (44) greater than in hospital/clinic I have sufficient knowledge of commu244 (59) nity services to personally plan/deliver home health care Home environment/family support is 256 (66) usually adequate for care at home All home care agencies offer same qual117 (32) ity/range of services I am satisfied with service quality of 319 (87) home care providers in my area Physicians should use home care agencies 330 (87) more Non-MD home care services adequately 109 (41) covered by third party payors With readily available home nursing (RN, 230 (57) aide), most MD house calls unnecessary 269 (67) I am too busy with office/hospital practice to make house calls P < 0.05 comparing characteristic across the three referral frequency groups. ** P < 0.01 comparing characteristic across the three referral frequency groups. *** P < 0.005 comparing characteristic across the three referral frequency groups. # Percents shown in parentheses are column percents.
Intermediate (12-47 Referrals) (n = 417) (%I
High (>47 Referrals) (n = 315) (%)
206 (51)
181 (60)*
373 (90)
278 (89)**
185 (45)
145 (46)
216 (56)
155 (53)*
309 (74)
263 (85)***
288 (73)
225 (74)
129 (33)
84 (29)
366 (88)
284 (89)
324 (84)
240 (83)
116 (41)
103 (46)
238 (59)
183 (60)
249 (60)
197 (63)
1246
BOLING ET AL
IAGS-DECEMBER 1992-VOL. 40, NO. 12
TABLE 5. HOME HEALTH AGENCY REFERRAL FREQUENCY ACCORDING TO PHYSICIANS' STRATEGY FOR CARE OF ACUTE PROBLEMS IN HOMEBOUND PATIENTS** Annual Home Care Referral Frequency Low (0-11Referrals) Physician's Strategy
(n = 429) (%)
Intermediate (12-47 Referrals) (n = 417) (%)
High (>47 Referrals) (n = 315) (%)
Given a homebound patient with an acute, non-emergent problem would you:* Send patient to hospital ER by ambu73 (18) 56 (14) 43 (14) lance Transport patient to office by ambu83 (20) 76 (18) 52 (17) lanceIvan Schedule physician house call 53 (13) 49 (12) 32 (10) Send office nurse, NP, or PA on 9 (2) 17 (4) 11 (4) house call Schedule a home care nursing visit by 56 (14) 75 (18) 102 (33) an outside agency Schedule a home care nursing visit 11 (3) 12 (3) 3 (1) and MD house call Other (eg, phone management) 122 (30) 126 (31) 70 (22) *- P < 0.001 cornparing tl7e proportion of physicians wi70 select a given strategy across the three referral frequent!/ groups by chi-square analysis
(7
x 3 tabld. '"I;
Pcrcerifagcs iii parentheses are coluniii percents.
TABLE 6. LINEAR REGRESSION MODEL OF HOME HEALTH AGENCY REFERRAL FREQUENCY WITH PHYSICIAN AND PRACTICE CHARACTERISTICS# Standardized Variable Beta Coefficient Intercept Greater hours per week employed by medical school Physician belief that malpractice risk is not increased in home care vs other setting Physician specialty, internist Greater physician age Physician sex, female Greater physician hours of patient care per week Home health agency nursing service is easily available Physician has served as home health agency director, board member, or consultant Physician comfort with own knowledge of community resources Physician graduation from North American medical school Rural practice location Higher percentage of geriatric patients in practice Greater number of house calls per year Greater number of homebound patients
0.000 -0.002 0.003 0.016 0.012 0.018 0.025 0.045 0.067* 0.078* 0.082* 0.096** 0.1 63***
0.133*** 0.156***
R ' = 0.126. * P < 0.05 probability of statistically significant independent contribution to model. *:* P < 0.01 probability of statistically significant independent contribution to model. **:l P < 0.001 probabilit!y of statistically significant independent contribution to niodel. # Negative beta co-efficient indicates a negative effect of the predictor variable on home health agency referral frequency.
ability of nursing services did not independently predict referrals. Using more extreme definitions of "rural" status in the model did not alter these findings.
DISCUSSION According to the 1987 National Medical Expenditure S ~ r v e y 5.9 , ~ million Americans used some home care services, altogether costing $12.1 billion. The Brooking Institution has projected that total home health care expenditures, measured in constant dollars, will triple in the next three decades.I3 Since physician oversight
is required for most home care services, physician involvement has fiscal relevance as well as clinical importance for the community-based health care team. In this national survey of family physicians and internists, most respondents reported some involvement with home health agency care. While 37 percent of respondents made less than 12 annual home care referrals, the typical physician reported making about three referrals per month, and 27 percent of respondents made frequent referrals, averaging 95 per year. These data reflect both the extent of the community
JAGS-DECEMBER 1992-VOL. 40, NO. 12
need and the reliance of physicians on home health agencies. Considered in fiscal terms, given that one typical episode of home health agency care involves 25 visits, at a cost of $66 per visit,I4 the average physician would annually order almost $60,000 worth of home health agency services. This study helps define some determinants of physician home care referral practice. Physicians’ personal and professional characteristics are often related to their practice patterns. For example, physician house call fre uency has been associated with physician specialty’ and with physician age.” However, in this survey, home care referral frequency was not related to most personal or professional characteristics of the physicians. Home care referrals showed a stronger association with practice profiles. The higher referral frequency for physicians with more work hours may reflect busier clinical practices. The larger proportion of geriatric patients and number of homebound patients among those making more referrals was also expected. Likewise, it was not surprising to find that those 168 (14 percent) respondents who had worked as a home health agency medical director, consultant, or member of a home health agency board of directors made more referrals. Often physicians selected for this role have extensive home care experience. Practical constraints precluded evaluating physicians’ vested interests as a motive. The finding, confirmed by the regression model, that home care referrals were less common among physicians graduating from medical schools outside of North America was unexpected. This might reflect unfamiliarity with community resources or culturally determined practice patterns. However, there are many possible interpretations, and further investigation would be required to further explore this issue. Rural physicians reported more home care referrals, and rural location was a strong independent predictor in the regression model. While service unavailability and logistic problems might tend to limit rural referrals, other practice demands and distances between homes might increase referrals, substituting for direct physician service in rural locations. Although nursing care was equally available to rural and non-rural physicians, other services such as physical therapy, social work, and appliances (eg, wheelchairs) were far less often considered readily available in rural areas. Some of these services are provided by part-time contract professionals who may be less available to the nonprofit agencies, which are often the only available agencies in rural areas. Overall, lack of service availability may be a significant problem in rural settings where home care appears to be a particularly important strategy for care delivery. Most of the physicians endorsed a wide variety of indications for home health agency referrals. However, compared with those who made fewer home care referrals, physicians making frequent referrals were much more likely to use home health agencies in many situations. Thus physicians who frequently use home health agencies may be more comfortable with an expanded health care delivery role for home health
HOME HEALTH AGENCY REFERRALS
1247
agencies. There were large differences in referral fre-. quency related to attitudes reflecting chronic care, such as the need to assess the home, improving compliance, providing terminal care, and management of chronic medical problems. Moreover, home health agencies were more often used in response to urgent care situations by physicians reporting more frequent referrals. Thus the tendency of physicians who make more frequent referrals to use home health agencies more actively appears to span the spectrum of acute and chronic care. Although almost 90 percent of respondents were satisfied with the quality of the available services, less than one-third of respondents considered the quality of home health agency services to be uniform. However, despite the variation in perceived service quality, referral frequency was not related to perceived service quality. There was no association of referral frequency with the type of agency available (eg, for-profit or hospital-based). Nor was referral frequency related to attitudes reflecting the physicians’ freedom from other responsibilities to make house calls or the belief that home health agency services may substitute for physician house calls. This finding is consonant with the observation that referral frequency increased with house call frequency, suggesting that home health agency care is not simply being substituted for physician care of the homebound. Concern about inadequate family support might also affect physician use of home health agencies. Although less than half of the physicians felt that most families can learn complex caregiving tasks, two-thirds considered the home environment adequate for home care, and referral frequency did not vary with these attitudes. Overall, more than 90 percent of physicians agreed that most families can learn to perform care at home. Three attitudes suggest a relationship between physician knowledge and home care referral practice. These are: the physicians’ belief that they can provide adequate medical care in the home; physician comfort with their knowledge of community resources; and belief that there is no increase in malpractice risk when ordering home care. The importance of physicians’ selfperceived knowledge of community resources was confirmed by the regression model. Whether physician home care knowledge is a result of active home care practice, or the practice a consequence of education, cannot be resolved by these data. The data did not support the hypothesis that availability of home health agency services drives referrals. Although nursing services and homemaker services were more often easily available to physicians making more referrals, the differences were modest. Most service categories were considered available by most respondents. Conversely, 15 percent of respondents did not consider nursing services readily available, a finding of potential concern considering the extent of the community need. Not surprisingly, use of the various individual home health services (eg, physical therapy or social work) was consistently greater among physicians making
1248
IAGS-DECEMBER 1992-VOL. 40, NO. 12
BOLING ET AL
more referrals. Use of the full spectrum of home health agency services may reflect recognition of the needs of the homebound by home health agency personnel. Often, home health agency personnel develop the home care plan of treatment, because many physicians are insufficiently familiar with home care regulations or patients’ home care needs to write treatment plans.’* The physician’s role in overseeing home health agency care deserves special comment. Third-party regulations require written physician approval for most home care services, and home health agency professionals often rely on physicians for telephone consultation during the course of treatment. Although time spent on case management did not increase linearly with home health agency referral frequency in this study, there was a consistent increase in time spent as referral frequency increased. Survey respondents making large numbers of referrals reported spending nearly 5 of 65 hours worked per week in activity related to the supervision of home care. The average physician spent 2.1 hours per week on telephone supervision of home care, and another 1.5 hours per week on related paperwork. This problem is compounded by new Medicare regulations which increased the requirement for direct physician involvement in completing forms authorizing use of durable medical equipment.I6 Despite the recent creation of charge codes specific to home care case management, Medicare intermediaries refuse to pay case management charges. Thus, while consuming substantial time, most physician work in support of home health agency care is un-reimbursed, undermining future physician motivation. The linear regression model places the major findings of this survey in perspective. Availability of nursing services, concern about malpractice, and physician work hours did not independently predict home health referral frequency. Moreover, physician specialty did not predict referral frequency, despite the fact that family medicine residency curriculum guidelines place greater emphasis on home care.17 On the other hand, variables reflecting the number of homebound patients in the physicians‘ practices were the strongest predictors of referrals, suggesting that patient needs play a substantial role in driving home health referrals by physicians. The importance of home care in rural practice is also confirmed by controlling for availability of nursing services and other factors. Finally, physicians’ self-perceived knowledge of community resources independently predicts home care referrals, suggesting an important educational agenda. Some limitations of this study should be noted. Although the response rate is comparable to that in most physician surveys,’8,l9 the possibility of response bias must be recognized. We could not evaluate attitudinal differences, but we did evaluate demographic variables and adjusted for the few differences between the respondents and the study population. Also, inherent in any survey reporting attitudes is the possibility that attitudes evolve to match behavior and that changes in attitudes will not necessarily predict changed behaviors. However, the respondents were clearly selective in their endorsement of attitudes.
Moreover, the attitudes are consonant with those previously reported in smaller samples.”, Third, we cannot verify the reported physician home health agency referral frequency and related time involvement. Finally, concern about response rate and cost forced a number of difficult choices during the survey design. The goal was to study both home visits and home care referral practice is broadly as possible. Thus, we could not insert many more detailed questions about such policy-relevant but complicated issues as the intermittency of the Medicare skilled home care benefit, differences between uses of homemakers and personal care aides, different levels of family caregiver sophistication, or physicians’ home care experience during training. The findings are also limited to family physicians and internists. We cannot comment on the home health agency referral practice of osteopaths, pediatricians, surgeons, psychiatrists, general practitioners, or obstetrician-gynecologists. However, family physicians and internists provide most of adult primary care, so we believe the study describes the majority of adult home care referral practice. Due to the length of the survey, data regarding house calls and “high tech” home care, such as home intravenous therapy, are being reported separately. Featured predominantly in the explosive growth of home care are professional services provided by home health agencies under physician supervision. This survey shows that most internists and family physicians make a large number of home health agency referrals and spend an appreciable amount of uncompensated time in work related to this referral practice. This finding is of particular importance, since relative work is now being measured and valued while reimbursement is increasingly being bundled. A substantial minority of physicians have less active home care referral practices, which may relate to several factors, such as lack of knowledge about home care or lack of homebound patients. Since home care is a key part of our society’s response to the needs of its rapidly increasing homebound population, these findings indicate a need for policy reform and educational initiatives to encourage continuation and enhancement of the physician‘s role in care of the homebound.
REFERENCES 1. Rowland D, Lyons
B. Triple jeopardy: Rural, poor, and uninsured. HSR
1989;23:975-1004. 2. Stone RI, Murtaugh CM. The elderly population with chronic functional
disability: Implications for home care eligibility. Gerontologist 1990;30: 491-496. 3. Committee on Ways and Means, U.S.House of Representatives. In: The 1991 Green Book. Washington, DC: U.S. Government Printing Office, 1991, p 150. 4. Spiegel AD. The growth of homecare. In: Spiegel AD. Home Health Care, 2nd Ed. Owings Mills, M D National Health Publishing, 1987, pp 329366. 5. Short PF, Leon J. Use of home and community services by persons ages 65 and older with functional difficulties. (DHSS Publication No. (PHS) 90-3466). National Medical Expenditure Survey Research Findings 5, Agency for Health Care Policy and Research. Rockville, MD. Public Health Service. 6. Branch LG, Wetle TT, Scherr PA et al. A prospective study of incident comprehensive medical home care use among the elderly. Am j Public Health 1988;78:255-259. 7. Council on Scientific Affairs of the American Medical Association. Home care in the 1990s. JAMA 1990;263:1241-1244.
IAGS-DECEMBER 1992-VOL. 40, NO. 12 8. American College of Physicians. Home health care. Ann Intern Med 1986; 105:484-490. 9. Public Policy Committee of the American Geriatrics Society. Home care and home care reimbursement. J Am Geriatr SOC1989;37:1065-1066. 10. Keenan JM, Hepburn KW, Ripsin CM et al. Home care agencies perceptions of Minnesota physician home care practice and skills: Results of a statewide survey. Fam Med 1992;24:142-144. 11. Boling PA, Retchin SM, Ellis J et al. The influence of physician specialty on housecalls. Arch Intern Med 1990;150:2333-2337. 12. Keenan JM, Bland CJ, Webster L, Myers S. The home care practice and attitudes of Minnesota family physicians. J Am Geriatr Soc 1991;39: 1100-1 104. 13. Rivlin A, Wiener J. Caring for the Disabled Elderly, Who Will Pay? Washington, D.C.: The Brookings Institution, 1988, p 42.
HOME HEALTH AGENCY REFERRALS
1249
14. National Association for Home Care. Basic Statistics About Home Care1991. Washington, DC, p 2. 15. Schueler MS, Hams DL, Goodenough GK et al. House calls in Utah. West J Med 1987;147:92-94. 16. Medicare Carriers Manual. Health Care Finance Administration Publication no. 6, Sections 60-64. Revised August, 1989. 17. Directory of Graduate Medical Education Programs. Section 2: Requirements for accreditation of programs. Chicago, IL: American Medical Association, 1989, pp 26-27. 18. Gunn WJ, Rhodes IM. Physician response rates to a telephone survey: Effects of monetary incentive on response level. Pub1 Opinion Q 1981; 45:109-115. 19. Babbie ER. Survey Research Methods. Belmont, CA: Wadsworth Publishing CO. Inc., 1973, p 165.
