Author Affiliations: Center for Studying Health System Change, Washington, DC.
Communication between primary care physicians (PCPs) and specialists regarding referrals and consultations is often inadequate, with negative consequences for patients. We examined PCPs' and specialists' perceptions of communication regarding referrals and consultations. We then identified practice characteristics associated with reported communication.
We analyzed the nationally representative 2008 Center for Studying Health System Change Health Tracking Physician Survey of 4720 physicians providing at least 20 hours per week of direct patient care. Outcome measures were physician reports of communication regarding referrals and consultations.
Perceptions of communication regarding referrals and consultations differed. For example, 69.3% of PCPs reported “always” or “most of the time” sending notification of a patient's history and reason for consultation to specialists, but only 34.8% of specialists said they “always” or “most of the time” received such notification. Similarly, 80.6% of specialists said they “always” or “most of the time” send consultation results to the referring PCP, but only 62.2% of PCPs said they received such information. Physicians who did not receive timely communication regarding referrals and consultations were more likely to report that their ability to provide high-quality care was threatened. The 3 practice characteristics associated with PCPs and specialists reporting communication regarding referrals and consultations were “adequate” visit time with patients, receipt of quality reports regarding patients with chronic conditions, and nurse support for monitoring patients with chronic conditions.
These modifiable practice supports associated with communication between PCPs and specialists can help inform the ways that resources are focused to improve care coordination.
Effective communication between primary care physicians (PCPs) and specialists regarding patient referrals and consultations is necessary for coordinated care,1 is important to patients and physicians,2- 5 and improves patient outcomes and physician satisfaction.6,7 Interspecialty communication is increasingly important because medical subspecialization and technological advances fragment care across numerous physicians.1,8,9
Despite its importance, interspecialty communication occurs inconsistently.6,10- 13 Identification of practice factors with the potential to enhance communication can inform efforts to improve care coordination. Such efforts include health information technology (HIT) adoption; patient-centered medical homes, which emphasize coordination by PCPs; and accountable care organizations (ACOs), which strive for shared accountability for patients among physicians.14- 18
Others have examined interspecialty communication in specific settings or situations and found, for example, that use of standardized note formats and computer access to medical records were associated with better communication.2,6,10,11,13 In another study,19 practices in which a care coordinator worked on-site with the patient's PCP for patients with chronic conditions had improved coordination and were cost neutral.
We examined a nationally representative physician sample to describe PCP and specialist perceptions of communication regarding referrals and consultations. We then identified practice characteristics associated with positive perceptions of interspecialty communication. We hypothesized that greater practice resources (including fewer time pressures), the presence of care management supports, and indicators of more established referral networks and integration would be associated with greater communication regarding referrals and consultations. We were particularly interested in 3 modifiable factors: adequacy of patient visit time (as a proxy for time for communication and practice resources), use of HIT to communicate regarding patients, and use of nonphysician staff to support care management.
We analyzed a national stratified random sample of 4720 physicians from the 2008 Health System Change Health Tracking Physician Survey. Funded by the Robert Wood Johnson Foundation, it is the fifth in a series of nationally representative physician surveys conducted since 1996 by the Center for Studying Health System Change.20 The sampling frame included active licensed allopathic and osteopathic physicians and was drawn from the American Medical Association master file. The mail survey achieved a 61.9% response rate. Characteristics of physicians who completed the survey and those who refused to participate were similar.20 All survey items have been validated through successful use in other surveys or cognitive testing.
Respondents must have completed their medical training and provide direct patient care for at least 20 hours per week. Physicians with minimal direct patient contact (eg, radiologists, anesthesiologists, and pathologists), federal employees, and temporarily licensed foreign medical graduates were ineligible. Additional background information is available at http://www.hschange.org/CONTENT/1085/. An institutional review board approved the survey and procedures.
The PCPs included family practice, general practice, general internal medicine, geriatric, internal medicine–pediatric, and general pediatric physicians. Before combining all specialists within 1 group, we analyzed several subgroups separately (eg, types of surgeons with differing needs to communicate with PCPs and cognitive vs procedural specialists).
Care coordination requires that PCPs referring patients to specialists communicate timely, relevant patient information and the reason for the referral and that specialists communicate findings and recommendations back to the PCP.21 Equally important is the receipt and recognition of this information by the respective parties.1 On the basis of primary care and chronic care frameworks,1,22 we examined factors hypothesized to influence these components of communication.
We used 4 questions adapted from the Primary Care Assessment Tool, Provider Survey.23 We asked PCPs, “When referring a patient to a specialist, how often do you send the specialist notification of the patient's history and the reason for consultation?” and “How often do you receive useful information about your referred patients from specialists?” We asked specialists, “When you see a patient referred to you by a PCP, how often do you receive notification of the patient's medical history and reason for consultation?” and “For patients who were referred to you by a PCP, how often do you send the PCP notification of the results of your consultation and advice to the patient?” Response options were “always,” “most of the time,” “sometimes,” “seldom or never,” and “not applicable.”
Reports of communication receipt reflect other physicians' sending behavior and the receiving organization's ability to get communications to the appropriate physician when needed. Thus, results regarding communication receipt may reflect the characteristics of physicians in the referral network and those of the receiving practice.
Independent variables of primary interest were organized conceptually into the 3 areas related to our hypotheses. We recognize that some (eg, practice type and size) might fit into all 3 areas.
We asked physicians about the extent to which they thought they had adequate time to spend with their patients during office visits (Likert scale responses ranging from “agree strongly” to “disagree strongly”) and the practices' payer mix (percentage of Medicare and Medicaid revenue). Because practices caring for patients with more complex and comorbid conditions face greater coordination challenges, we assessed whether the practice cared for patients with 1 or more of 4 prevalent and costly chronic conditions (eg, asthma, diabetes mellitus, depression, and congestive heart failure) and the percentage of the physician's patients with any chronic condition.
Other independent variables relevant to practice resources and physician time pressures were practice type and size and physician base compensation method. We also included physician reports of the importance of quality measures, patient satisfaction surveys, productivity, practice profiling, and practice financial performance in determining compensation (moderately or very important vs somewhat or not important).
Practice supports for patients with chronic conditions indicate a capacity for care management and are hypothesized to be associated with communication in connection with consultations and referrals. These supports include (1) nonphysician patient educators, (2) nurse care managers, and (3) the receipt by the physician of quality reports regarding his/her patients with chronic conditions.22- 24 These items were asked of physicians in practices treating any 1 of 4 prevalent chronic conditions (eg, asthma, diabetes, congestive heart failure, or depression). In the logistic regression models, a dummy variable was included to capture physicians in practices that did not provide care for patients with any of these 4 conditions.
We also assessed the presence and use of HIT in the practice, focusing on elements with the potential to support communication and information sharing. After assessing them individually, we created a summary measure of 3 HIT tools most relevant to coordination: (1) whether the respondent's main practice uses an electronic medical record (EMR) (all paper, 0; part electronic, 1; and all electronic, 2); (2) whether the physician routinely used HIT for accessing patient notes, problem lists, and medication lists (no, 0; yes, 1); and (3) whether the physician routinely used HIT to exchange clinical data with other physicians (no, 0; yes, 1). We summed responses and then categorized them as high (3-4), partial (1-2), or none (0). The rate of e-mail use for communication with other physicians regarding patients was too low to be included in the analysis.
We included proxy variables that helped capture the extent to which physicians have more vs less established referral networks with other physicians: physician years in practice (<11, 11-20, >20), based on prior findings that physicians in practice longer have more established working relationships with other physicians; rural vs urban location; independent physician ownership; practice type (group or staff model health maintenance organization [HMO] vs community health center vs other practice types) and group practice size (number of physicians); number of managed care contracts; and percentage of practice revenue under capitation.
We controlled for additional physician factors, including sex, board certification, medical training location (United States or Canada vs elsewhere), and specialty.1,2,6,25 We also controlled for geographic region, physicians' assessment of “market competitiveness,” and county ratio of PCP-to-specialist supply.
We examined univariate frequencies of all variables, correlations among independent variables, and the bivariate associations between independent and outcome variables. Tests for multiple comparisons of the bivariate proportions were conducted, and they did not differ significantly from the raw P values. On the basis of these analyses and the clinical and policy relevance of particular factors within the primary care and chronic care conceptual frameworks, we built logistic regression models. The goal of our multivariate logistic regressions was to examine the association between the key modifiable independent variables (eg, adequate time with patients) and each binary outcome (eg, sends information regarding a referral or consultation), adjusting for all other practice factors and potential confounders.
A separate logistic regression model was created for each of the 4 coordination outcome measures. Two of the models applied to PCPs and 2 applied to specialists. All 4 contained the same independent variables. To estimate the independent effects of physician, panel, and market characteristics, we entered nonmodifiable factors (physician sex; years in practice; board certification; location of training; practice type; percentage of patients in practice with chronic conditions; percentage of practice revenue from Medicare, Medicaid, and capitation; region; and urban status) into each logistic regression model as mutually exclusive categorical variables. After building these base models, variables that could theoretically affect communication and coordination of care were added one at a time (eg, revenue sources, factors affecting compensation, use of an EMR, use of nurse care managers for patients with chronic conditions, and whether the physician thought he or she had adequate time to spend with patients during visits). To assess the relative contributions of the key independent variables of interest on the outcomes and model fit, we performed the likelihood ratio test (χ2 difference in the likelihood ratios) as each modifiable factor was added to the base model.
We assessed for interactions between key independent variables and relevant covariates with respect to the coordination measures; none were significant. Specialist subgroup results were consistent, with only minor differences in magnitude but not in direction of the associations; therefore, only combined specialist results are given.
We determined present adjusted percentages from final parsimonious models. Hosmer-Lemeshow tests indicated good fit of the data. Also, SUDAAN software, version 10.0 (Research Triangle Institute, Research Triangle Park, North Carolina) was used to analyze the data that were weighted for probability of selection and survey nonresponse. Unadjusted analyses were largely consistent with the multivariate logistic regression results and thus are not presented.
Finally, we calculated the predicted increase in interspecialty communication as each of the 4 key practice supports (HIT use, adequate visit time, quality reports regarding patients with chronic conditions, and a nurse care manager) is added for an average respondent, holding all other covariates at their means. These simulations are based on estimates from the multivariate logistic regression models.
Table 1 presents the sample characteristics. Primary care physicians are more likely than specialists to think that they do not have adequate time to spend with patients during office visits (39.3% vs 30.2%, P < .001). Specialists are significantly more likely to have no managed care contracts (14.1%) compared with PCPs (9.7%, P < .001). A higher percentage of specialists compared with PCPs have high HIT use (38.4% vs 35.5%, P = .05). A much higher percentage of PCPs (64.7%) receive reports on the quality of care for their patients with chronic conditions than do specialists (25.4%, P ≤ .001).
The PCPs' and specialists' reports of the extent that they communicate with one another regarding patient referrals and consultations differed significantly (Table 2). Although 69.3% of PCPs reported that they “always” or “most of the time” send the specialist notification of a patient's history and reason for consultation at the time of referral, only 34.8% of specialists reported they “always” or “most of the time” receive such notification. Similarly, 80.6% of specialists said they “always” or “most of the time” send the referring PCP notification of the results of their consultation and advice to patients, whereas only 62.2% of PCPs reported they received such information. The PCPs and specialists who reported not consistently receiving communications regarding referrals and consultations were significantly more likely to report that their ability to provide high-quality care was threatened because of failure to receive timely reports (Table 3). This finding suggests poor communication threatens quality of care and supports our outcome measures' content validity.
For PCPs and specialists, “having adequate time to spend with patients during the office visit” was the factor with the greatest positive association with each communication outcome (Table 4 and Table 5). Although physicians with higher proportions of patients with chronic conditions were more likely to report sending notification to their colleagues, they were also more likely to report that receipt of timely reports from other physicians was a problem (data not shown).
Receipt of quality reports regarding patients with chronic conditions was positively associated with 3 of 4 outcomes (Table 4 and Table 5). The involvement of a nurse at the practice to help treat patients with chronic conditions was associated with greater receipt of communication regarding referrals and consultations. Use of HIT was associated with higher reports of receiving and sending communication by specialists but not by PCPs.
Physicians who have practiced 20 years or more and physicians in small and nonmetropolitan areas report sending and receiving communication regarding referrals and consultations at higher rates than their counterparts, regardless of specialty. The PCPs in group- and staff-model HMOs, community health centers, and hospital or medical school practices report sending referral information to specialists at higher rates than counterparts in group practices (Table 4). Among PCPs, reports of sending communication to specialists decreased as the number of managed care contracts increased (Table 4). For specialists, this association was in the opposite direction.
No significant associations were found between “the importance of factors affecting compensation” (eg, quality measures) and the outcomes. A rerun of models excluding physicians in institutional settings (eg, staff-model HMOs and hospitals) did not change these results. Physician income, practice ownership, board certification, and county ratio of primary care to specialist physicians were also not consistently associated with physicians' communication (data not shown).
The Figure illustrates the predicted increase in interspecialty communication as each of the 4 key practice supports (HIT use, adequate visit time, quality reports regarding patients with chronic conditions, and a nurse care manager) is added for an average respondent. For example, PCPs with all 4 practice supports would increase their reported sending of referral communication from 63.9% to 82.7%, relative to PCPs with none of the 4 supports.
Percentage of physicians reporting they “always” or “most of the time” send or receive communication regarding referrals and consultations at baseline and then with the addition of each of the 4 practice supports. HIT indicates health information technology; PCP, primary care physician.
To our knowledge, ours is the first nationally representative study of physicians to describe interspecialty communication regarding consultations and referrals. The results suggest much room for improvement. The PCPs and specialists reporting inconsistent receipt of communications from other physicians were significantly more likely to report that their ability to provide high-quality care was threatened because of failure to receive timely reports.
The significant differences in PCPs' and specialists' perceptions of the extent to which they communicate with one another regarding referrals and consultations may have various causes, including overstatement of sending behaviors and understatement of receipt. Although our questions asked specifically about patients referred by PCPs, specialists may have misattributed some patient self-referrals as PCP referrals. In addition, process failures, such as late or misdirected reports and reports not placed in the patient's record in a timely fashion at the recipient practice, may affect responses regarding receipt despite timely sending.
The factor most consistently and strongly associated with interspecialty communication was “adequate time with a patient during the office visit.” In addition, PCPs were more likely to believe that they do not have adequate time to spend with patients during office visits than were specialists. This likely occurred because of the complex nature of primary care, which involves longitudinal care over time and coordination of a patient's care across various conditions.1 Increased administrative burden (eg, health plan prior authorizations) and decreasing reimbursements create pressures for physicians to see more patients26 and decrease the time available for effective communications regarding referrals and consultations during and after the visit. The average primary care face-to-face visit lasts 10.7 to 18.7 minutes, depending on the assessment method,27,28 and time pressures during the visit result in lower-quality care, including poorer patient and physician care experiences.29,30 Our findings complement these prior studies. With more encounter time, the physician and staff can focus more completely on the patient's needs, which may include initiating referral and consultation communication and retrieving reports from other physicians.
Overall, payment factors, as measured by this survey, had inconsistent and little association with interspecialty communication. Most current compensation methods do not incentivize communication. Practice supports may be more important to facilitating interspecialty communication, at least in the predominant fee-for-service environment.
The positive association between receipt of quality reports regarding patients with chronic conditions and interspecialty communication regarding referrals and consultations suggests the potential benefits of providing performance feedback to physicians. Such feedback may help identify gaps in care and spur physicians to communicate with others sharing care for patients. Practice environments where quality reports are generated may also have cultures or mechanisms to facilitate referral and consultation reports or follow-up when communications are not received.
It is notable that significantly fewer specialists received quality reports compared with PCPs for their patients with chronic conditions. Given the positive association between quality report receipt and communication levels between PCPs and specialists, increased attention to providing specialists with such quality reports might be indicated. In addition, incorporating interphysician communication measures into such reports might help improve coordination.
The association between having a nurse in the practice help coordinate care for patients with chronic conditions and greater receipt of interspecialty communication may reflect efforts by the nurse to reach out to other practices to obtain referral or consultation letters. This is a particularly important function for overwhelmed practices, although few have the resources to provide nurse care managers and few payers compensate for this. Even in primary care practices that lack nurse care managers, the presence of nonphysician staff to educate patients with chronic conditions was beneficially associated with the sending of reports from one practice to another.
The positive association between HIT use and sending or receiving information among specialists, but not among PCPs, may reflect that specialists are more likely to work in large or institutional settings with available HIT.31 Use of HIT in this national sample was higher than prior estimates,32 which used different measures of HIT use, but closer to a recent publication of data from the National Ambulatory Medical Care Survey, which found that 41.5% of physicians reported using all or partial EMR systems.33 Outside large integrated health care systems and large multispecialty groups, however, lack of interoperability, current EMR design, and limited clinical processes and reimbursement for coordination limit the extent to which present-day EMRs are used for communication between primary care and specialist practices.34
The PCPs with greater administrative burden, as represented by more managed care contracts, reported lower sending behaviors. Multiple contracts also likely indicate a broader set of physician networks in which a physician participates and, consequently, less well-defined referral networks.9 Not surprisingly, physicians in more integrated settings, such as group or staff-model HMOs, were also significantly more likely to report communication regarding referrals and consultations.
This study's strengths include the nationally representative sample, including physicians in small and medium groups that constitute most outpatient practices in the United States27 and that have received less attention in the quality improvement literature. The survey also had a high response rate for a physician survey, and we controlled for a wide range of physician, practice, and market characteristics.
The study's limitations include the use of physician self-report data that, given the study's national scope, could not feasibly be validated by medical record reviews. We also lack information on the accuracy, comprehensiveness, and timing of communication. However, our national sample's self-reported rates of sending or receiving communication regarding referrals and consultations are within the range reported in the literature.2,6,10- 13 Regardless, potential self-report bias (owing to social desirability) should not systematically affect the direction of associations. Finally, the cross-sectional design does not permit causal attribution.
Although we are unable to fully capture the constructs of practice resources and pressures and referral networks and integration, the individual items examined can help inform policy and clinical initiatives by identifying how resources might be focused to improve communication regarding referrals and consultations. Our results suggest that supporting PCPs so that they, for example, can afford nurse care managers may facilitate adequate communication with specialists.
Efforts to improve coordination should address the low rates of interspecialty communication regarding referrals and consultations. Current initiatives such as patient-centered medical homes, HIT, and ACOs have the potential to strengthen financial incentives, structures, and care processes to support communication. Combined with prior work24,30,35 and anticipated lessons from current medical home experiments, our findings can help inform these initiatives, suggesting ways to focus resources to support infrastructure and measurement of interspecialty communication. Targeted support for capabilities identified in this study as being associated with good bilateral communication—adequate visit time with patients, quality reports regarding patients with chronic conditions, and nurse support for coordination—may help advance the communication that is critical to care coordination and the success of policy efforts to improve it. Such input might also support larger organizational efforts; for example, ACOs may help to define a tight referral network and shared accountability for patients, whereas the patient-centered medical home within an ACO can focus on care management and serve as the hub of coordination efforts.
In conclusion, systematic structures, tools, and processes for information creation,36 transfer, receipt, and recognition by the sending and receiving physicians are needed to assist medical practices. Measures of “meaningful HIT use”14 and coordination of care could include items that support, track, and confirm completion of each of these tasks.
Correspondence: Ann S. O’Malley, MD, MPH, Center for Studying Health System Change, 600 Maryland Ave SW, Ste 550, Washington, DC 20024-2512 (firstname.lastname@example.org).
Accepted for Publication: June 5, 2010.
Author Contributions: Both authors had access to all the data in the study and take full responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: O’Malley and Reschovsky. Acquisition of data: O’Malley and Reschovsky. Analysis and interpretation of data: O’Malley and Reschovsky. Drafting of the manuscript: O’Malley and Reschovsky. Critical revision of the manuscript for important intellectual content: O’Malley and Reschovsky. Statistical analysis: O’Malley and Reschovsky. Obtained funding: O’Malley. Administrative, technical, and material support: O’Malley. Study supervision: O’Malley.
Financial Disclosure: None reported.
Funding/Support: This study and the 2008 Health System Change Health Tracking Physician Survey were funded by the Robert Wood Johnson Foundation.
Additional Contributions: James E. Ward, JD, edited an early version of the manuscript.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Internal Medicine editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 19
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.