To evaluate the effect of medical comanagement on outcomes of hospitalized surgical patients who had postoperative complications.

 We used clinical and administrative data at a large urban hospital to conduct a cohort study of select surgical hospitalizations in 2008 and 2009. We identified patients who suffered postoperative complications using measures developed by the University Health System Consortium. Bivariate and multivariate regression analyses were used to determine the associations of postoperative comanagement with inpatient mortality, length of stay, and cost in surgical patients who had postoperative complications.

 From 21,728 total surgical hospitalizations, we identified 4040 hospitalizations involving primary procedures (mainly orthopedic and neurosurgical) that were associated with comanagement at least 25% of the time. After excluding cases with missing data, 501 hospitalizations (13.8%) involved a patient who suffered at least 1 postoperative complication. Patient characteristics between the comanaged (n = 297) and non-comanaged (n = 204) hospitalizations were well matched. Medical comanagement was associated with fewer in-hospital deaths (odds ratio 0.23, 95% confidence interval 0.05-0.99) in adjusted analysis. Comanaged compared with non-comanaged hospitalizations were associated with shorter stay (—2.6 days, 

 <.01) without significant differences in total cost.

 Comanagement of patients who had perioperative complications was associated with lower mortality, suggesting that comanagement may facilitate effective rescue among medically complex surgical patients.

	We evaluated the effect of medical comanagement on outcomes of hospitalized surgical patients who had postoperative complications.

		 Comanagement of patients who had perioperative complications was associated with lower mortality.

		 The higher cost of comanaged hospitalizations mandates development of tools that allow for the cost-effective application of comanagement.

	A recent study analyzing a large cross-sectional data set from the National Surgical Quality Improvement Project found that postsurgical complications—many of them acute medical conditions&mdash;occur at similar rates among hospitals regardless of inpatient mortality rates.

 This finding suggests that the best hospitals distinguish themselves less by preventing complications than by their apparent ability to 

 patients who have suffered complications during hospitalization. This and other studies

 support the view that many adverse events in surgical patients are due to problems in medical management rather than intraoperative care, and suggest the possibility of reducing adverse outcomes in this population by improving care outside of the operating room.

	Medical comanagement has been adopted by many surgeons as a potential method to improve the quality of postoperative care.5 Comanagement involves a medical generalist or specialist who manages patient issues that arise on the wards, allowing surgeons to dedicate their time to their specialtyspecific activities. Comanagement can be understood as a patient referral wherein the comanaging physician participates in hospital care broadly, in contrast to traditional consultations directed toward addressing particular problems.

 This practice arrangement has rapidly been adopted over the last decade, such that about half of all surgical hospitalizations in the Medicare population are now comanaged.

 Although recent studies have shown that focused preoperative consultations by medicine physicians do little to prevent postprocedural complications and mortality,

the effects of postoperative comanagement on clinical outcomes remain uncertain.

	To understand the impact of postoperative comanagement on the care of hospitalized surgical patients, we analyzed data from our academic hospital’s electronic medical records. We hypothesized that comanagement of hospitalized surgical patients would have a protective effect on risk-adjusted mortality through the rescue phenomenon suggested by the National Surgical Quality Improvement Project data,1 as well as by other studies.

To this end, we examined comanagement’s associations with inpatient death among patients who suffered postsurgical complications. We also assessed the impact of comanagement on risk-adjusted length of stay and hospital cost.

	We conducted our study at an urban 854-bed teaching hospital. Surgeons on faculty belonged to 1 of 17 surgical specialties. Surgical services were staffed by an attending physician with house staff and medical students. Inpatient services of neurosurgery, urology, and orthopedic, transplant, and vascular surgeries were also staffed by nonphysician providers.

	The perioperative medicine service was staffed by 7 general internists from the Division of Hospital Medicine who provided preoperative evaluation and postoperative comanagement services. Comanagement was available 7 days a week during daytime hours. Infrequently, cross-covering night hospitalists performed new urgent consultations during evening hours with handoffs of care to the perioperative physicians in the morning. Some high-risk patients seen in the preoperative medical clinic, including all high-risk spine surgery patients, were comanaged postoperatively at the internists’ discretion. Surgeons also requested comanagement for additional patients

	based on perceived need. Postoperative involvement by the perioperative physicians is referred to as comanagement in this study because of their exclusive focus on surgical patients, their orientation toward improving postsurgical care quality, and the higher intensity of involvement in their patients’ care reflected by the typically greater frequency of encounters compared with other medical specialty consultants.

	All inpatient adult surgical discharges following a procedure in 2008 and 2009 were identified using the institution’s Enterprise Data Warehouse, which contains all data from the hospital’s electronic administrative and medical records. The use of comanagement varied by individual surgeons’ practice, which led to nonuniform adoption of comanagement in the hospital. We sought to reduce heterogeneity in the population by matching patients primarily on surgical procedures. Using

 International Classification of Disease, 9th Revision

) procedure codes, we selected hospitalizations with primary procedures associated with comanagement at least 25% of the time. Patient demographic and clinical information such as the American Society of Anesthesiologists (ASA) physical status score,

 data to calculate the Charlson Comorbidity Score,

and hospital cost data were obtained. We also obtained information about the total number of nonsurgical consultants who were involved in each hospitalization. Comanagement was defined as hospitalizations in which at least 1 perioperative internist documented an encounter during the postoperative course.

	Postoperative complications for our patient population were identified by cross-linking to the University Health System Consortium (UHC) database.

 Member institutions submit surgical discharge abstract information to the consortium, which then uses diagnosis-related groups and 

 codes to identify rates of 24 specific surgical complications. The institutional review board of Northwestern University approved the study protocols, consents, and data collection mechanisms.

	We compared patient demographic and risk characteristics using the

 test for categorical and continuous data, respectively. Postoperative complications were tabulated and differences in frequency by comanagement status were compared using the X

	Bivariate analyses were performed to assess associations between inpatient death and potential risk factors. We fit a mixed-effects logistic regression model to determine the adjusted association between comanagement and inpatient mortality, allowing for clustering around individual surgical procedure. Unadjusted median costs and lengths of hospitalization by comanagement status were compared using the Wilcoxon rank-sum test. Risk-adjusted comparisons were performed using generalized linear regression modeling.

 shows a flow diagram of the study subjects. Of  the 1048 primary procedures accounting for 21,728 surgical hospitalizations during the study period, 184 primary procedures were found to be associated with comanagement at least 25% of the time. Of 4040 hospitalizations, 420 before March

	2008 were excluded because ASA scores were not electronically documented. Of the remaining 3620 hospitalizations, 501 (13.8%) involved patients who had at least 1 postoperative complication and were included in the analysis.

Characteristics of Patients and Their Hospitalizations

	Characteristics of patients and their hospitalizations are shown in 

. The 3 subgroups of admission source were ambulatory recovery, referring to patients whose hospitalization followed surgery in an ambulatory setting; emergency department; and inpatient unit, referring to patients transferred from nonsurgical services or from other hospitals. More comanaged than non-comanaged hospitalizations followed procedures performed in the ambulatory setting. Fewer procedures in the comanaged group were designated emergencies. The mean duration of surgery was significantly longer for comanaged hospitalizations.

	An orthopedic procedure was the primary surgery in the largest number of hospitalizations (n = 280, 55.9%), while spine surgery (n = 128, 25.6%) and neurological surgery (n =39, 7.8%) were next most commonly involved. The top 6 procedures accounted for 87% of the hospitalizations in the study and included total knee replacement (

 codes 81.03-81.08, 81.34, 81.35), hip replacement (

 codes 70, 72, 73, 81.51, 81.5), reduction of the femur (

lists the UHC postoperative complication measures in the study’s hospitalizations in descending order of frequency. Deep vein thrombosis or pulmonary embolism was the most common complication (31.9%). Hospitalizations involving pneumonia and other infectious complications were more likely to be comanaged.

Association Between Comanagement and Survival

	Bivariate analyses found significant positive associations between inpatient mortality and the following variables in descending order of magnitude: ASA score greater than or equal to 4, emergency surgery status, inpatient unit admission source, the number of nonsurgical consultants, and the Charlson Comorbidity Index (

). Mixed-effects regression analysis with clustering around procedures found no statistical significance in the associations between inpatient death and emergency surgery status and admission source. Moreover, comanagement was associated with lower risk-adjusted mortality (odds ratio [OR] 0.23, 

 = .049) during the postoperative hospitalization.

	When using unadjusted comparisons, we found that both median length of hospitalization and cost were higher in comanaged compared with non-comanaged hospitalizations (

). Risk adjustment, however, yielded shorter mean length of hospitalization by 2.6 days (

<.01) among comanaged compared with non-comanaged hospitalizations. The risk-adjusted cost of hospitalization was not different by comanagement status.

	In this exploratory observational study, postoperative comanagement by a general internist focused on perioperative care was associated with lower risk-adjusted mortality among surgical patients who had a postoperative complication. Comanaged hospitalizations were associated with shorter riskadjusted lengths of stay with no difference in risk-adjusted cost. Due to the traditionally low mortality rate in the surgical inpatient population, the effect of interventions to further lower mortality in this population is difficult to demonstrate. Although care quality interventions such as the implementation of comanagement may not be expected to have a dramatic impact on mortality, our finding is intriguing and generates a new hypothesis around the failure-to-rescue phenomenon.

	The focus of this study on the effects of postoperative comanagement is a departure from several recent studies that examined the effects of preoperative consultations.

Although results are not shown, our data confirm the absence of a mortality benefit with preoperative consultations, which suggests that preventing many postoperative complications is difficult. Our analysis, however, suggests that the widespread adoption of postoperative comanagement may find its benefit in the role of hospitalists as facilitators of rescue. The exact nature of this contribution from comanaging hospitalists was not explored in this study, but may be related to their disease-specific experience in perioperative medicine,

 their availability to respond to emergencies on the wards similarly to rapid response teams,

 and surveillance for emerging signs of complications, as seen with skilled floor nurses and house staff.

 Moreover, hospitalists’ clinical focus on process improvement activities may have salutary influences on hospital care quality that have been negatively associated with failure to rescue.

	The impact of hospitalist comanagement of a neurosurgical service was explored recently by Auerbach and colleagues using a retrospective interrupted time-series analysis.

 Although the study by Auerbach et al did not detect a decrease in adjusted mortality with the implementation of comanagement, a careful comparison of patient characteristics found that comanaged patients in the study were older and had more medical problems than their non-comanaged counterparts. Comanaging hospitalists in this setting may be playing an important supportive role in facilitating care for an increasingly elderly and comorbid surgical population. Although the mortality benefit of comanagement remains hypothetical, future studies may focus on the impact of comanagement on the most vulnerable surgical patients.

	Comanagement is not implemented without added cost. Comanagement of neurosurgical hospitalizations has been associated with a small risk-adjusted cost savings.

This savings was not replicated in our study, and the higher unadjusted cost may reflect extra testing and treatment administered by hospitalists. Because comanagement benefits a relatively small group of at-risk patients, we see the need for clinical criteria that can guide its targeted implementation to the appropriate patient population.

	A few observations are noteworthy from our regression analysis. First, our findings support a previously tested assertion that risk prediction scores derived from administrative data such as the Charlson Comorbidity Index are poorer predictors of mortality than clinical instruments like the ASA score.

 Not surprisingly, the number of nonsurgical consultants involved in hospitalizations appeared to be a surrogate indicator of clinical complexity, but no other covariate was significantly associated with the outcome in adjusted analysis.We found that the association between comanagement and mortality became stronger and statistically significant only after adjusting for other covariates, particularly the number of nonsurgical consultants. We believe that a possible explanation involves selection bias resulting from some surgeons’ pattern of consulting multiple specialists (at the exclusion of comanaging hospitalists) for patients who manifest early signs of complications. An alternative explanation is that surveillance for complications is enhanced under comanagement compared with specialist consultations. Due to the inherent limitations of administrative data, we were unable to adjust for the timing and severity of complications, resulting in residual confounding.

	The study is limited in several other ways. The major limitation of this observational study is that comanagement was not randomly assigned. We selectively limited our analysis to a group of patients with similar characteristics. The exposure groups varied with respect to admission source, there were fewer emergency surgeries in the comanaged group, and the mean operating room time in the comanaged group was longer due to the preponderance of high-risk complex spine surgeries. Although none of these were independent predictors of mortality, we acknowledge the potential for omitted variables and residual confounding in all observational studies. As with any

Articles

WEB EXCLUSIVE

Medical comanagement of patients who had perioperative complications was associated with lower mortality, suggesting that comanagement may facilitate effective rescue.

Potential Role of Comanagement in Rescue of Surgical Patients

To evaluate patient and physician characteristics associated with the purchase of antidepressant medication for at least 6 months.

Clalit Health Services is the largest managed care health fund in Israel, a country that uses a universal healthcare system. We randomly sampled 30,000 primary care patients over the age of 22 years. Our analytic sample consisted of those 949 patients who did not purchase antidepressant medication during the last quarter of 2005 and purchased antidepressant medication at least once in 2006. We used multilevel analysis, with whether or not medication was purchased for at least 6 months as an indicator of adherence. Patient and physician characteristics were evaluated as potential predictors.

Only 23% of the sample was classified as adherent. Physician characteristics explained only a small portion of the variance in adherence and, as a result, were not included in multivariate analysis. Patients who did not have a somatic diagnosis, had a depression diagnosis, and were of higher socioeconomic status were more likely to be classified as adherent. Patients who purchased tricyclic antidepressants were less likely to be classified as adherent.

The findings suggest that in a managed care setting, there is high uniformity among physicians. Although physician characteristics explain little of the variability associated with adherence, certain patient characteristics as determined by their physicians (eg, antidepressant drug class, psychiatric diagnosis) do play a role in adherence.

	Characteristics of patients and physicians at Israel’s largest HMO were evaluated as potential predictors of antidepressant adherence (ie, the medication was purchased for 6 months or longer).

		 Only 23% of the patient sample was classified as adherent.

		Although physician characteristics explain little of the variability in adherence, certain patient characteristics as determined by their physicians (eg, antidepressant drug class, psychiatric diagnosis) do play a major role in adherence.

		Even in a country that has universal healthcare, population group differences exist and are associated with socioeconomic status.

	The public health costs associated with untreated or undertreated depression or anxiety are substantial.

Primary care providers serve as the main gatekeepers for mental health problems in Israel as in other countries,

 with both depression and anxiety being most often managed in primary care clinics by primary care providers.

Fortunately, the past few decades have seen a tremendous improvement in the management of depression and anxiety.

 According to current clinical guidelines, the first line of treatment for the management of either depression or anxiety should be an antidepressant such as a serotonin reuptake inhibitor (SSRI) or a serotonin-norepinephrine reuptake inhibitor. Older antidepressant medications, such as tricyclic antidepressants (TCAs) or monoamine oxidase inhibitors, may also be indicated for specific populations or as augmentation therapy.

 Clinical guidelines recommend the use of these agents for a minimum of 6 consecutive months in order to achieve adequate efficacy and stabilize the psychiatric condition.

	Despite the documented efficacy of these pharmacologic agents in the management of depression and anxiety, reports of nonadherence are staggering.

 Studies have shown that more than 30% of the patients discontinue their medications within the first 4 weeks of treatment1

 and more than 60% discontinue within the first 6 months.

	Reasons for nonadherence vary and may relate to patient, drug, and provider characteristics. Studies that have focused primarily on patient characteristics found that individuals of ethnic minority background, younger or older adults,

 individuals of lower socioeconomic status (SES),

and those with a personality, substance use,

 somatization, depression, or anxiety diagnosis

 are less likely to adhere to their regime. In contrast, others found that carrying a dual depression and anxiety diagnosis results in better adherence.

 Beliefs concerning psychotropic medications,

 or a mismatch between patients’ preferred and assigned treatment

 have also been shown to be barriers to adherence. In addition, the side effect profile of a particular drug26 is of importance, with certain psychotropic drugs resulting in lower adherence than others.

	It has been argued that physicians’ knowledge about appropriate treatments, and their ability to establish rapport, assess patients’ attitudes, and provide appropriate consultation and information are all essential for establishing appropriate adherence.

 Hence, as expected, physician characteristics have also shown to be associated with adherence to antidepressant medication. Whereas some found that receiving services from providers with a mental health specialty results in better adherence to psychotropic medication,

others noted that simply providing patients with educational messages concerning the importance of the medication improves adherence.

 was also associated with improved adherence. Nevertheless, several different interventions specifically geared toward increasing adherence to psychotropic medication were found to be largely ineffective,

 leading researchers to argue for a need for a substantial shift in services at the organizational level in order to improve adherence.

 Thus, questions have arisen about the provider role in improving adherence to psychotropicmedications.

	Although informative, the majority of research to date focused either on patient characteristics or on provider characteristics as facilitators of adherence, but did not take both into account simultaneously. In addition, results are still equivocal with regard to the role of primary care physicians as facilitators of adherence to antidepressant medications. Moreover, the majority of research was conducted in countries that do not use a universal healthcare system. Thus, although we know that the prohibitive costs of psychotropic medications may be a potential barrier for some patients,

 we do not know whether disparities in adherence continue in the absence of financial constraints to the patient.

	The present study provided a unique opportunity to evaluate patient and provider characteristics associated with adherence to antidepressant medications in primary care clinics in Israel, a country that uses a universal healthcare system, providing a comprehensive, uniform basket of services that includes psychotropic medications, but not psychotherapy.

 We used a comprehensive computerized medical registry of Clalit Health Services, the largest healthcare fund in Israel, which insures 53% of the Israeli population and operates as a managed care organization. Under the Israeli system, the primary care provider is considered responsible for all patient healthcare needs. The provider is responsible for the care of a defined number of patients and refers patients for consults as needed. Based on expert recommendations, the primary care provider determines patients’ treatment. The database includes information on patient characteristics as well as primary care provider characteristics, thus allowing a distinctive analysis of the actual purchasing behavior of patients instead of relying on self-reports, which have known limitations, in particular regarding sensitive issues.

	Based on past research concerning nonadherence to psychotropic medications, we expected both patient and physician characteristics to be associated with nonadherence. Specifically, we expected older and younger individuals as well as ethnic minorities to show higher levels of nonadherence. We also expected those who did not carry a psychiatric diagnosis to have lower levels of adherence. We expected physician specialty to be associated with adherence (ie, patients of physicians who specialize in family medicine would demonstrate higher levels of adherence). This expectation was based on past research, which has shown that family medicine physicians are more likely than internal medicine physicians to manage depression by themselves.

	We randomly sampled 30,000 primary care patients over the age of 22 years as of January 2006 from Clalit Health Services’ computerized medical registry. Of these patients, 591 died in 2006 and 4446 had incomplete physician data. Thus, the source sample consisted of 25,086 primary care patients. (These are not mutually exclusive; one can have missing values and be dead at same time.) Of these, 1219 had a documented purchase of antidepressant medication in the last quarter of 2005 and were excluded from analysis. Our study group consisted of the 949 patients who did not purchase antidepressant medication during the last quarter of 2005, but did purchase antidepressant medications at least once in 2006.

	Clalit Health Services uses a computerized medical registry that contains both patient and physician data. All physician visits are recorded in this registry and are matched with pharmacy data using patient and physician unique IDs. For the purpose of the study, data were deidentified to prevent patients’ breach of confidentiality.

Adherence to Antidepressant Medications. 

We extracted data regarding antidepressant (eg, SSRIs, TCAs, other) purchase. Adherence to a new antidepressant regime was defined as no purchases of antidepressant medication in the last quarter of 2005, followed by at least 6 months of purchase starting in 2006. Although purchasing is not necessarily synonymous with adherence as indicated by actual usage of the medication, it does indicate an active decision on the part of the patient and involves a financial investment. In addition, once antidepressants are purchased for 3 months or longer, they are considered a chronic medication by Clalit Health Services. As such, they are listed in all patient-related correspondence. According to the clinical guidelines of Clalit Health Services, discontinuation of a chronic medication is not automatic, but instead requires a discussion between patient and provider.

 Age, sex, birth country, and number of years in Israel are listed in the registry based on patients’ self-report during their first visit.

The Charlson Comorbidity Index (CCI) score uses both the number and seriousness of medical conditions to predict mortality. The CCI includes 19 medical conditions that are weighted on a scale of 1 to 6 based on their 12-month relative mortality risk, with total scores ranging from 0 to 37. Because age was found to be an independent risk factor for mortality, one score is added to the CCI total score for each decade of life over the age of 50 years.

 Past research has recommended the use of the CCI as a continuous measure rather than a categorical one, in order to refine its ability to distinguish between patients.

 This score is calculated by Clalit Health Services’ medical division for all primary care patients, adjusted for their age.

A list of common nonpsychiatric somatic conditions that may indicate a psychiatric condition was constructed by the authors based on the medical literature.

These conditions include the various types of arthritis pain, head and back pain, abdominal pain, and insomnia. The presence of any of these diagnoses was classified as a somatic diagnosis.

Mood disorders, anxiety disorders, adjustment disorders, and somatoform disorders were included under this category, using the

 International Classification of Diseases, 10th Edition

, codes. We excluded post-traumatic stress disorder, obsessive compulsive disorder, bipolar disorder, and body dysmorphic disorder, as our aim was to focus on psychiatric conditions that are within the mandate of Clalit’s primary care physicians,

Medications were classified into 4 groups: SSRIs, TCAs, other, or a combination of at least 2 of these classes.

Socioeconomic status (low, medium, high) was based on the Israeli Central Bureau of Statistics classification.

 Although these data are available at the primary care clinic level and not at the patient level, they are considered a good proxy of patient SES.

 The vast majority of patients are registered in neighborhood clinics; thus, the SES of clinics reflects that of the patients’ place of residence.

Individuals were classified as Arabs or Jews based on Clalit Health Services’ classification of primary care clinics into locations that serve at least 70% Israeli Jews versus Israeli Arabs. Although these data are available at the primary care clinic level, they are considered a good proxy of patients’ ethnicity, given Israel’s highly segregated nature.

	Physician age, sex, place of birth, number of years in the country, number of years of experience, and specialty (none, family, other) were available from employee records.

	We first conducted univariate and bivariate analyses. Next, we conducted multilevel analysis, with patient-level data representing the first level of predictors (eg, patient age, sex, SES) and physician-level data (eg, physician age, sex, number of years in the country) representing the second level of predictors. The outcome variable was whether or not antidepressant medications were purchased for at least 6 months starting in 2006 with no purchases in the last quarter of 2005.

	The first step of a multilevel analysis used an empty model with physician random effect. This model estimated the outcome per physician rather than per patient. This analysis yielded an intraclass correlation (ICC) score, which ranged between 0% and 100%. The ICC reflected the degree to which patients of the same physician were more similar to one another than to patients of other physicians. Thus, it reflected the proportion of the total variance that was due to differences between physicians. If the ICC is relatively large, a multilevel analysis is justified. On the other hand, if the ICC is relatively low, a multilevel analysis is unjustified, and analysis should take into consideration only patientlevel variables (ie, level 1). As a rule of thumb, ICCs of .05, .10, and .15 represent small, medium, and large effect sizes, respectively.

	Given the binary nature of the outcome, we used the linear threshold model to obtain an ICC.

 Following the calculation of ICC (which was less than 1%), logistic analysis was conducted with the purchase of antidepressant medications for at least 6 months as an outcome. Only significant predictors identified in bivariate analyses were entered into the model.

	The final sample consisted primarily of women (63%); most patients were over the age of 46 years (72%). The sample consisted primarily of Jews born in Israel (34.8%) or in East Europe (24%). Most of the sample had no psychiatric diagnosis (69.3%) and most purchased SSRIs (57%). The majority of physicians were female (54.8%) and were born in East Europe or the former Soviet Union (54.8%). Most had a specialty in family medicine (51.4%). See

	Only 23% of the sample was classified as adherent. In bivariate analysis, those purchasing for 6 months or longer were less likely to have a somatic diagnosis, more likely to have a psychiatric diagnosis, and more likely to purchase SSRIs or combined psychotropics. They also were more likely to be of higher SES. See Table 1.

	Results of logistic regression analysis revealed that those who did not have a somatic diagnosis, had a diagnosis of depression, and were of higher SES were more likely to be classified as adherent. In addition, compared with those purchasing SSRIs, those purchasing TCAs were less likely to be classified as adherent. See 

	The present study evaluates patient and provider characteristics associated with adherence to antidepressant medications. This study is unique for several reasons. First, we evaluated patient and provider characteristics simultaneously in an attempt to identify their exclusive roles in adherence. Second, we used a representative sample of patients of the largest managed care organization in Israel, a country that uses a universal healthcare system, thus eliminating financial barriers to patients. As a result, we could evaluate disparities, which may stem from reasons other than patients’ financial status. Third, we evaluated purchasing data rather than self-report data concerning nonadherence, and thus were more likely to get an accurate estimation of nonadherence. Finally, the focus on all patients who purchased antidepressant medications rather than only on those diagnosed with depression or anxiety was valuable, given the tendency to underdocument mental illness in primary care.

	The most notable finding of the present study is that only 23% of patients purchasing antidepressant medications were classified as adherent. This figure likely represents an underestimate of nonadherence, given the fact that patients who were prescribed antidepressant medications, but for some reason did not purchase the medications even once, were not included in our analysis. Findings suggest that even in a country that uses a universal healthcare system, nonadherence is substantial.

Patient characteristics such as psychiatric diagnosis were associated with variations in adherence, although physician characteristics were not.

Disparities in Antidepressant Adherence in Primary Care: Report From Israel

To study risk factors for and likelihood of unplanned hospital admission and readmission in persons aged 

We used data from clinical records of the primary care centers, pharmacy database, and hospital discharge records for persons aged 

65 years registered in primary care centers of referral hospitals in the Baix Llobregat healthcare area. Unplanned hospital admission was defined as any unscheduled hospitalization in 2008; unplanned readmission was defined as any unscheduled admission occurring within 6 months after discharge date of index admission. Logistic regression analysis was used to identify predictors of unplanned hospital admission and readmission.

The population included 28,430 individuals. Among them, 2103 (7%) experienced an unplanned admission and 365 (1.3%) an unplanned readmission. The readmission rate for the admitted population was 18.7%. The strongest predictive factor of unplanned admission was >2 admissions in the previous 2 years (odds ratio [OR] 24.9, 95% confidence interval [CI] 16.0-38.7 for 2007; OR 15.6, 95% CI 8.6-28.0 for 2006). Factors associated with unplanned readmission were aged 

2 admissions during the previous year (OR 20.4, 95% CI 14.1-29.5). The area under the receiver operating characteristics curve was 0.78 for unplanned admission and 0.85 for unplanned readmission in the development sample and 0.76 and 0.81, respectively, in the validation sample.

 Aged persons and those who used more hospital services in previous years had a higher probability of hospital admission and readmission.

	Predictive models of people at high risk for hospital admission and readmission were developed using primary care, pharmacy, and hospital care databases.

		Aged people with respiratory and/or cardiac comorbidities and with greater use of hospital care services in previous years had a higher probability of hospital admission and readmission.

		 Application of this predictive model to identify people at risk may have a considerable impact on the care of patients with high comorbidity, making for a more proactive healthcare system and at the same time reducing the cost of avoidable hospitalizations.

	Persons of advanced age have a greater number of chronic diseases and more comorbidity than the remaining population. This population has a considerable economic impact on the healthcare system, particularly when there is a lack of coordination between primary care and hospital services. Patients with chronic diseases and high comorbidity may not always receive optimum care because sometimes they have not been identified as persons at risk or with special needs. Some studies have estimated that the population with greater needs represents around 10% of the total, but consumes about 70% of available healthcare resources.

 Unplanned hospital admissions and readmissions are included among the indicators of this higher healthcare expenditure. Short periods of time between admission and readmission (eg, 30 days) are used as indicators of quality of hospital care. Longer periods are useful as indicators of the population at risk and to asses the impact on the health system.

	Managed care models have been developed to promote integration and coordination of the healthcare system, and to provide patients with different interventions according to their level of need.

These models have been implemented by various organizations.

One of these models stratifies the population’s healthcare needs into 4 levels and provides patients with different interventions according to their level of need, while the other model focuses on the high-risk population and offers specific resources, particularly nursing case management, to achieve a more proactive approach in the care of high-risk patients. The aim of this strategy is to improve the patient’s clinical outcome and quality of life, while reducing the use of more costly services such as unplanned hospital admission. These models have also been applied in the National Health Service in England. Managed care models are based on identification of patients at risk of emergency hospitalization or high consumption of resources, or of requiring highcost care in the future.

	A systematic review published in 1991 identified 44 studies related to predictive models of hospital admission or readmission.

 Forty-six studies published between 1991 and 2010 that developed predictive models to identify persons at risk were retrieved; most of them cohort studies (n = 30). Eight standardized instruments

 designed to recognize the population at higher risk and to develop interventions to reduce unplanned admissions have been published.

	The healthcare system in Catalonia is a public system, financed by taxes, with universal healthcare coverage and free entry for all citizens. It is oriented toward primary care, which is the first contact and level of access users have with the healthcare system (cure, care, prevention, and health promotion), and it is mainly provided in primary care centers.

The hospitalization rate in Catalonia was 306.3/1000 for the population over age 64 years in 2009.

 A comparable figure of admission rate in acute, short-stay hospitals was 350.8/1000 for the United States in 2006.

	With the currently available information about the Spanish healthcare system, it would be appropriate to incorporate tools to help identify subpopulations at risk of high consumption of resources.

	The aims of this study were to analyze the risk factors for and the likelihood of unplanned hospital admission and readmission in the population over age 64 years who were registered with a primary care center in 1 of the geographical healthcare areas of Catalonia. This tool could potentially provide useful information to establish interventions focused on preventing future unplanned admissions and readmissions.

	This is a retrospective cohort study including all persons more than 64 years of age who were registered in the primary care area of Baix Llobregat Litoral within the health region of Costa de Ponent in Catalonia (Spain). The health services provided in this area include 7 primary care teams and 2 public referral hospitals, Viladecans Hospital, a first-level center, and Bellvitge Hospital, a third-level center for more complex cases. The primary care centers and hospitals provide almost all healthcare services in this area.

	The data were obtained from the e-CAP, the computerized information system used in primary care; the pharmacy database, which records dispensation of drugs; and the hospital discharge data from Viladecans and Bellvitge. Data used to build the predictive models came from the period encompassing 2006, 2007, and 2008. Data from 2006 and 2007 were used as background information on healthcare service use in both the primary care and hospital settings. To validate the model, data from the period including 2007, 2008, and 2009, were used, with data from 2007 and 2008 being the background information.

	The dependent variables were unplanned hospital admission between January 1, 2008, and December 31, 2008, and unplanned readmission within 6 months after discharge day of the index admission.

	The exclusion criteria for index admissions were elective or planned admission, readmissions during the same study period, admission of patients who resided outside the area, and hospital discharges without recording the admission or discharge date or the type of admission. In readmissions, the exclusion criteria were persons dead at discharge of index admission and individuals transferred from one hospital to another hospital in the same readmission.

	The sociodemographic factors analyzed were age (categorized into 65-69, 70-74, 75-79, and 

80 years), sex, place of residence, morbidity, and diagnoses recorded in the e-CAP, based on the 

International Classification of Diseases, 10th Revision, Clinical Modification

 into large groups, from which the more specific, common diagnoses according to the literature were chosen. These included insulin-dependent diabetes (E10), non—insulindependent diabetes (E11), heart failure (I50), ischemic heart disease (I20-I50), bronchitis (J40, J41, and J42), emphysema (J43), chronic obstructive pulmonary disease (COPD) (J44), asthma (J45), bronchiectasis (J47), Alzheimer’s disease (G30), and femoral neck fracture (S72), plus the total number of concurrent diagnoses (

5). In addition, the following factors were analyzed: consumption of primary care resources in the previous 2 years, consultations by telephone contact (0 vs 

4), and use of hospital resources in the previous 2 years, including previous unplanned admissions (

9). Categorization of these variables was carried out in keeping with data in the literature and according to their skewed distribution in the population studied.

	An analysis of the internal validity, quality, and consistency of the information obtained was carried out following a previously designed protocol.

	This study was performed in accordance with national and international guidelines (ethics code, Declaration of Helsinki) and with legal regulations on the confidentiality of personal data (Law 15/1999 of December 13 on the protection of personal data).

	Logistic regression was used to identify factors predictive of unplanned hospital admission in a 12-month period and readmission within 6 months. The model building process consisted of 3 steps: selecting the variables, building the model, and validating the model. Unplanned hospital admission and readmission models were adjusted for sex and age. Model fitting was realized with an iterative process, selecting in each step the candidate variable to add to the model, using a statistical significance level of <.01. To choose the best model, the likelihood ratio test was used to compare the models in each iteration.

	Validation of the model was performed by testing calibration, discrimination, and colinearity, and analyzing the residuals and influence values.

SPS version 18 (SPS Inc, Chicago, Illinois) and STATA version 11 (StataCorp, College Station, Texas) for Windows were used for the statistical analyses.

	A total of 28,430 persons over 64 years of age (57.6% women) were registered in a primary care center in the study area. Among them, 30.3% were 80 years or older, 35.7% were registered in Castelldefels, and 35.9% were registered in Viladecans, Spain.

	The sociodemographic characteristics of the entire study population, patients who had any hospital admission (planned or unplanned), and patients who had only an unplanned hospital admission in 2008, are presented in 

. Among the total population, 7.4% (n = 2103) had an unplanned admission; 51.7% of this group were women and 48% were 80 years or older (

). The most prevalent diagnoses related to unplanned admissions compared with the total population were COPD, heart failure, ischemic heart disease, non—insulin-dependent diabetes, and having 5 or more concurrent diagnoses (

). Compared with the total population, hospitalized patients in this group consumed an increased amount of healthcare resources, and this increase was higher in 2007 than in 2006.

	Among the total population, 1.3% (n = 365) had an unplanned readmission within 180 days from the index admission (53.7% men, 53.7% 80 years or older), and 4.1% had any readmission (unplanned or planned; 

). In the admitted population the rate of readmission was 18.7% (365/1952) at 6 months and 13.6% (265/1952) at 30 days. Compared with the total population, readmitted patients had a higher prevalence of almost all the variables analyzed. Emergency admission in the 2 years prior to readmission was also higher in the readmitted population (

	The predictive factors for unplanned hospital admission were sex (male), age (the older the patient, the higher the risk), COPD, heart failure, 5 or more concurrent diagnoses, dispensation of 4 or more medications, and the use of hospital resources in the 2 years prior to admission (emergency admissions, planned admissions, and cumulative days of hospital stay). Associated factors for unplanned readmission were sex (male), age (the older the patient, the higher the risk), insulin- dependent diabetes, ischemic heart disease, heart failure, the number of dispensed drugs, and emergency admissions and days of hospital stay accumulated in the previous year. The most powerful predictive factor for both admission and readmission was unplanned admissions in the previous year (odds ratio [OR] 24.9, 95% confidence interval [CI] 16.0-38.7 and OR 20.4, 95% CI 14.1-29.5, respectively; 

	The area under the receiver operating characteristic (ROC) curve of hospital admission was 0.78 in the study sample and 0.76 in the validation sample (

). The variance influence factors were 1.03 to 1.24 and the Hosmer-Lemeshow 

 value was .50. For readmission, ROC curves were 0.85 in the study sample and 0.81 in the validation sample (Figure). The variance influence factors were 1.00 to 1.14, and the Hosmer-Lemeshow 

	In this study, predictive models for unplanned hospital admission and readmission were developed in a health area of Catalonia using retrospective data from the primary care, pharmacy, and hospital databases. Hospitalization in previous years was the most powerful risk factor for unplanned hospitalization and rehospitalization. These predictive models enable identification of persons at higher risk and establishment of proactive interventions with greater intensity in the highrisk categories, so that more effective allocation of resources to meet population needs can be achieved.

	The fact that the Spanish healthcare system is oriented toward primary care, with easy and universal access to care, could be associated with low rates of hospitalizations, both the total rates and specifically the rates for ambulatory care—sensitive conditions.

 Nevertheless, factors related to th use of primary care did not achieve statistical significance in our predictive models. As almost all the elderly population had consulted regularly with primary care in previous years (mean number of visits to a primary care center per year in the study sample was 10.8 in 2008), these variables lost their ability to discriminate between those who were admitted and those who were not admitted to the hospital. Moreover, in spite of the high mean number of visits to a primary care center in the Spanish elderly population compared with the number of visits in other countries, the present study did not include assessment of the “quality” or effectiveness of those visits.

	No differences were found in the probability of admission or readmission related to the place of residence, despite the existing variability in clinical practice in Spain.

 This may be because the area analyzed was relatively small, including only 2 hospitals and 7 primary care teams.

	The rates of unplanned readmission for the admitted population in our study were 18.7% at 6 months and 13.6% at 30 days, both lower than the 20.0% generally found in the United States at 30 days.

	Previous studies in Spain have analyzed admission and/or readmission for specific diseases or specialties. In Catalonia (Baix Empordà) the relationship between population morbidity and healthcare costs was examined.

 In Valencia, relationships between comorbidity and hospital stay, mortality, and readmission were analyzed using the Charlson index.

The results from the current study cannot be compared with the results of these studies because we examined the risk of unplanned hospitalization using various databases, not using only hospital data. In addition, the study from Baix Empordà analyzed primary care and hospital cost together as a dependent variable.

Predictive models for hospital admissions and readmissions in persons aged >65 years have been developed through a combination of comorbidity and previous healthcare use.

Predictive Model for Emergency Hospital Admission and 6-Month Readmission

 To assess quality of depression care during the high-risk period following a psychiatric hospitalization.

 Retrospective administrative data analysis.

 Using Veterans Health Administration (VHA) administrative data, we assessed mental health follow-up within 7 and 30 days of psychiatric hospitalizations for major depression from 2004 to 2008. Adequate antidepressant medication coverage and number of psychotherapy visits were assessed within 90 days of discharge. Multivariable logistic regression was used to identify patient demographic and clinical characteristics associated with each quality indicator.

 Of the 45,587 patients discharged from a psychiatric inpatient stay with a diagnosis of major depressive disorder, 39.4% and 75.8% received an outpatient visit within 7 and 30 days of discharge, respectively; 58.7% of patients received adequate antidepressant coverage (72 of 90 days) and 12.9% received adequate psychotherapy encounters (8 visits). Receipt of outpatient mental health visits and of adequate psychotherapy were less likely among patients who were male, aged <35 or >65 years, had >3 major general medical comorbidities, lived >30 miles from a VHA clinic, or whose hospital length of stay was <7 days. Patients with comorbid substance use disorders were less likely to receive adequate antidepressant treatment.

 To optimize evidence-based depression care after a psychiatric hospitalization, health systems might increase receipt of psychotherapy by considering potential barriers related to age, medical condition, and distance. Patients with comorbid substance use disorders or their providers may need additional services to support antidepressant treatment.

	To enhance treatment of depression during the high-risk period following a psychiatric hospitalization, health systems should:

		Ensure timely access to outpatient mental health care and consider additional transition support, particularly for patients without mental health encounters prior to their psychiatric admission.

		Increase receipt of an adequate course of psychotherapy. Patients aged either <35 or >65 years, those with >3 serious medical comorbidities, and those residing >30 miles from a clinic were particularly less likely to receive psychotherapy.

		Increase antidepressant medication coverage for depressed patients with comorbid substance use disorders.

	Immediately following discharge from a psychiatric hospitalization, patients are at high risk for functional impairment, rehospitalization, and suicide.

Different patient characteristics predict adequate antidepressant treatment after hospitalization, received by 58.7% of patients, versus adequate psychotherapy, received by 12.9% of patients.

September 2011

Potential Role of Comanagement in Rescue of Surgical Patients

Disparities in Antidepressant Adherence in Primary Care: Report From Israel

Predictive Model for Emergency Hospital Admission and 6-Month Readmission

Depression Care Following Psychiatric Hospitalization in the Veterans Health Administration

Cost-Offset Analysis: Bimatoprost Versus Other Prostaglandin Analogues in Open-Angle Glaucoma

Quality Measurement of Medication Monitoring in the Meaningful Use Era

Electronic Health Records, Clinical Decision Support, and Blood Pressure Control

Compliance, Persistence, and Switching Patterns for ACE Inhibitors and ARBs

Mental Illness and Warfarin Use in Atrial Fibrillation

Inappropriate Antibiotic Prescribing in Managed Care Subjects With Influenza