Hospitalists, general internists and family physicians often serve as managers of hospital patients. This role has rapidly expanded for hospitalists in the US. In some instances the general physician has the lead role as attending physician seeking consultation with other specialist physicians. In other cases the specialist may be the attending physician and seek the additional support of the primary care physician or hospitalist. The benefit of this approach has been demonstrated in a few specific instances. Orthopedists and hospitalists co-managed patients with hip and knee arthroplasty with evidence of lower complication rates, but not lower costs or length of stay (LOS).1 In two studies,2,3 orthopedist and hospitalist care of hip fractures resulted in marginally beneficial cost and efficiency trends in one study2 and reduced LOS in the other study.3
While medical and surgical specialists may complement one another’s knowledge and skills, the same may not be equally true for general internal medicine physicians and internal medicine subspecialists, because the basic training of subspecialists includes general internal medicine. It is a common practice in hospitals throughout the US to have internal medicine physicians (general internists or hospitalists) act as principal attending physicians of patients who require medical subspecialty care. In these circumstances, the medical subspecialist is a consultant. The alternative situation arises when the medical subspecialist becomes the principal attending physician for the patient and may or may not ask a hospitalist or general internist to consult on the case.
Cardiovascular disease is the most common category of illness in hospitalized adult populations in the US. Relatively few studies have examined the efficiency or quality of hospital care as to whether a cardiologist or internist is the best principal attending physician for cardiovascular diseases and most studies have focused on heart failure only. Ahmed et al.4 compared cardiologists alone, generalists alone, and combined care for 1,075 hospitalized heart failure cases. Combined care resulted in better processes of care and lower rates of readmission. Cost of care was not evaluated and only heart failure diagnosis was included. Philbin et al.,5 in another study of 44,926 heart failure patients in New York, found that cardiologists and internists had similar LOS, charges, deaths, and readmissions and that family practice physicians had the lowest utilization profile.
The extent of consultation was not reported. Auerbach et al.,6 in the Study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT) of 1,298 hospitalized heart failure patients, reported that cardiologists’ patients incurred 43 % increased hospital cost compared with generalists’ patients, with a non-significant trend in adjusted mortality at one year favoring cardiologists’ care. In this study, 40 % of generalists’ patients had a cardiology consultation. Frances et al.,7 in a study of 161,558 elderly patients with myocardial infarction, reported a 2 % mortality reduction at one year for patients admitted to a cardiologist. Utilization was not studied, nor were consultation patterns by non-cardiologists. Finally, a study of care patterns in acute myocardial infarction comparing cardiologists alone, generalists, and generalists with cardiology consultation revealed more appropriate care for cardiologists alone or in consultation with generalists when compared with generalists alone.8 From the current literature it is reasonably clear that cardiologists alone or in consultation with generalists improve recommended care processes for heart failure and acute myocardial infarction. Mortality and cost patterns are inconsistent across studies. It is unclear from current studies if cardiologists should take the lead or should be a consultant to internists or hospitalists for the hospital care of cardiovascular diseases.
We sought to answer the following question for common cardiovascular diagnoses: does the cost and quality of inpatient care differ when a cardiologist is the principal attending physician compared with when an internist or hospitalist is the attending physician, with consultation by a cardiologist. A secondary question was whether procedure-based cardiovascular diagnoses would differ from non-procedure diagnoses with respect to the type of attending physician.
Methods
Setting. The study was carried out in a large, urban, not-for-profit community teaching hospital in Florida. The study began October 1, 2000 and ended June 30, 2005. These dates corresponded to the beginning of availability of severity categories in the data set and the end of an academic year. The study population consisted of all patients admitted to a general internist, internal medicine hospitalist, or cardiologist as the attending physician, but excluded all patients on a teaching service. The hospital supported residencies in internal medicine, medicine-pediatrics, pediatrics, general surgery, orthopedics, obstetrics and gynecology, emergency medicine, and pathology. There were no internal medicine fellowships.
Physician Groups—Community General Internists (Internists). One hundred three internists admitted patients to the hospital and served as attending physician. Nearly all were in solo or small group practices. These physicians organized their own night and weekend coverage arrangements but usually admitted their own patients and performed daily hospital rounds. These physicians usually practiced at more than one hospital in the city.
Hospitalists. Sixty-six hospitalists admitted patients to the hospital. They were organized into seven groups of varying size and none were employed by the hospital. Most of the patients were from local physicians who were not engaged in hospital practice. Each hospitalist group arranged coverage for nights and weekends and none was present in hospital 24 hours per day.
Cardiologists. Seventy cardiologists, organized in groups of varying size from solo practitioners to large group practices of more than 20 members, served as attending physicians for patients in the study group. Call coverage was arranged individually for each group practice, although an individual cardiologist was assigned each day for emergency department calls.
Data Source and Collection. Trendstar clinical costing software (McKesson HBOC, San Francisco, CA) was used to collect information on all hospitalized patients. Trendstar uses an activity- based cost accounting system derived from the hospital’s ledger.9,10 Costs are then reported including direct, indirect, fixed, and variable costs. Patients were grouped using All Patient Refined Diagnosis Related Groups (APRDRGs), severity level (1–4), and risk-of-mortality (ROM) level (1–4). APRDRG classification is assigned based on principal and secondary diagnosis, age, and procedure.11 The severity level and ROM are assigned within the APRDRG.11 All costs were assigned to the single attending physician of each admission.
Study Patients. All cardiovascular APRDRGs with more than 200 cases total and at least 50 cases per physician group during the study period were included. These high-frequency APRDRGs were selected in order to assure that statistical adjustment for confounding demographic factors, severity, and ROM could be carried out. There were 15,103 patients who met these criteria. The APRDRGs were subdivided into two groups that included a cardiovascular procedure performed by a cardiologist (Procedure group) or did not include a procedure (Without Procedure group).
All patients were admitted to the same hospital units and the intensive care and cardiac care units were of ‘open’ design. The attending physicians performed their own rounds and sought consultation as needed. The study was approved by the organization’s Institutional Review Board prior to the investigation.
Design and Statistical Analysis. The design was a retrospective cohort study. The hypothesis tested was that there was no difference among hospitalists, general internists, and cardiologists for hospital cost, LOS, hospital mortality, or 30-day readmissions, regardless of the principal attending physician. Endpoints (hospital cost, LOS, hospital mortality, and 30-day readmission) were calculated per hospital admission. Physician fees were not included in costs. Costs were controlled for inflation by introduction of an adjustment factor for year of study in the multivariate analysis, thus assuring that costs were compared equally over time. Readmissions within 30 days were attributed to the original discharging physician regardless of who subsequently admitted the patient.
The statistical analysis was performed using SAS 9.1 (SAS Institute, Cary, NC). Because of skewness and non-normality, costs and LOS were log-transformed prior to analysis. The highest and lowest 0.5 % cost admissions were removed as outliers. General linear modeling (GLM) was used to adjust for differences in confounding variables for cost and LOS endpoints. Logistic regression analysis was used to adjust for confounding variables for the endpoints of 30-day readmission and hospital mortality. Statistically significant independent variables were determined using stepwise automatic variable selection procedures. Age and physician group were always contained in the model. Statistical significance was set at p<0.05 for confounding variables to remain in the models. Pairwise comparisons within the models were analyzed by t-tests with Tukey’s adjustment for multiple comparisons.
Results
Table 1 provides basic descriptive information about the patient characteristics within demographic, severity, APRDRG, and physician categories. As expected, patient characteristics differed among physician groups and APRDRG categories. Generally, hospitalist patients were slightly younger and more frequently had commercial insurance compared with the cardiologist or internist patients. Cardiologist patients were more frequently Caucasian and of slightly lower severity category.
Table 2 displays the consultation patterns of the principal attending physicians for each of the APRDRG categories. The purpose of this part of the analysis was to determine the extent of cardiologist consultations by hospitalists and internists and to evaluate the use of hospitalists and internists and other consultants when the cardiologist was the principal attending physician. In order to assess consultation frequencies, two separate 5 % randomly selected samples of patient charts were reviewed and all consultations were logged. Table 2 represents the total of all information from the two samples. In general, the hospitalists and internists nearly always consulted a cardiologist in these APRDRGs. Also, cardiologists were very unlikely to consult hospitalists or internists when the cardiologist was the attending physician. Thus, the predominant utilization analysis compares cardiologists working without hospitalists or internists with hospitalists and internists working with a consulting cardiologist.
Table 3 reveals the LOS and hospital admission cost (Cost) by attending physician type and by APRDRG category after statistical adjustment for age, insurance, ethnicity, gender, APRDRG, and severity. A general linear model regression analysis within each APRDRG category was performed with Cost or LOS as the dependent variable and age, insurance, ethnicity, gender, APRDRG, severity, year, and physician category as independent variables. Cardiologists had statistically significantly lower overall hospital costs than hospitalists or internists for the diagnoses in the analysis, whether with or without a procedure. Overall LOS was also statistically significantly lower for cardiologists compared with internists, and also compared with hospitalists when a procedure was involved. The pattern of reduced Cost and LOS was consistent across lower and higher severity levels and was especially evident when the diagnosis included a cardiology procedure.
Table 4 summarizes the discharge status, hospital mortality, and 30-day readmission rates of the physician category for each diagnosis category. Mortality and 30-day readmission rates were analyzed using logistic regression analysis with age, gender, ethnicity, insurance, and physician category as independent variables. Severity was also an independent variable for 30-day readmission and risk of mortality was also an independent variable for mortality. Odds ratios with 95 % confidence intervals were produced. After adjustment for confounding variables, there were no statistically significant differences among the physician groups in hospital mortality or in 30-day readmissions.
Discussion
Our data suggest that, for the selected diagnosis groups of inpatients in our study, cardiologists working mostly alone as attending physician generally perform better economically than hospitalists or internists when acting as the principal attending physician of record.
Quality-of-care markers did not significantly differ among the physician categories after adjustment for patient characteristics. For cardiology procedure-related diagnoses, the Cost and LOS differences were especially evident. Our study expands the evidence that cardiologists can provide cost-effective inpatient care as the attending physician in cardiovascular diseases, when compared with an internist or hospitalist attending with consultation by a cardiologist. Our study is the first one that we are aware of to evaluate these attending physician models of care for several procedure and non-procedure cardiovascular diagnosis groups.
Our finding that hospital cost and LOS were more nearly comparable among the three physician categories for non-procedure diagnoses and more divergent for procedure diagnoses makes sense. When a cardiovascular procedure is needed, the cardiologist is free to schedule and complete the procedure without waiting for a consultation request from another physician. When no procedure is involved, our study and most other studies have found only modest differences in cost or quality between generalists and cardiologists. Our study is similar to most other studies in finding little or no difference in readmission or death rates.5–7,12 We did not evaluate processes of care or the use of guidelines by the physicians in our study. Our data and conclusions have limitations which we acknowledge. The data came from one hospital setting and may not be generalized to other settings. We used statistical procedures to adjust for multiple confounding variables that are of potential significance in making economic and quality-of-care comparisons. However, many other factors that we could not account for could have introduced a bias that would alternatively explain our results. Our study did not evaluate global healthcare costs.
It is possible that one group of physicians shifted more costs to the outpatient environment than other groups or performed more procedures than are necessary and thus increased overall costs rather than decreased them. We could not study this issue.
Finally, the optimal design to evaluate utilization and quality of care would be a randomized trial.