Hospital Readmission and Post-Acute Care Use After Intensive Care Unit Admissions: New ICU Quality Metrics?

OBJECTIVE: Care coordination is a national priority. Post-acute care use and hospital readmission appear to be common after critical illness. It is unknown whether specialty critical care units have different readmission rates and what these trends have been over time. METHODS: In this retrospective cohort study, a cohort of 53,539 medical/surgical patients who were treated in a critical care unit during their index admission were compared with 209,686 patients who were not treated in a critical care unit. The primary outcome was 30-day all cause hospital readmission. Secondary outcomes included post-acute care resource use and immediate readmission, defined as within 7 days of discharge. RESULTS: Compared to patients discharged after an index hospitalization without critical illness, surviving patients following ICU admission were not more likely to be rehospitalized within 30 days (15.8 vs. 16.1%, p = 0.08). However, they were more likely to receive post-acute care services (45.3% vs. 70.9%, p < 0.001) as well as be rehospitalized within 7 days (5.2 vs. 6.0%, p < 0.001). Post-acute care use and 30-day readmission rates varied by ICU type, the latter ranging from 11.7% after admission in a cardiothoracic critical care unit to 23.1% after admission in a medical critical care unit. 30-day readmission after ICU admission did not decline between 2010 and 2015 (p = 0.38). Readmission rates declined over time for 2 of 4 targeted conditions (heart failure and chronic obstructive pulmonary disease), but only when the hospitalization did not include ICU admission. CONCLUSIONS: Rehospitalization for survivors following ICU admission is common across all specialty critical care units. Post-acute care use is also common for this population of patients. Overall trends for readmission rates after critical illness did not change over time, and readmission reductions for targeted conditions were limited to hospitalizations that did not include an ICU admission.

Characteristics, Outcomes, and Trends of Patients With COVID-19-Related Critical Illness at a Learning Health System in the United States.

BACKGROUND: The coronavirus disease 2019 (COVID-19) pandemic continues to surge in the United States and globally. OBJECTIVE: To describe the epidemiology of COVID-19-related critical illness, including trends in outcomes and care delivery. DESIGN: Single-health system, multihospital retrospective cohort study. SETTING: 5 hospitals within the University of Pennsylvania Health System. PATIENTS: Adults with COVID-19-related critical illness who were admitted to an intensive care unit (ICU) with acute respiratory failure or shock during the initial surge of the pandemic. MEASUREMENTS: The primary exposure for outcomes and care delivery trend analyses was longitudinal time during the pandemic. The primary outcome was all-cause 28-day in-hospital mortality. Secondary outcomes were all-cause death at any time, receipt of mechanical ventilation (MV), and readmissions. RESULTS: Among 468 patients with COVID-19-related critical illness, 319 (68.2%) were treated with MV and 121 (25.9%) with vasopressors. Outcomes were notable for an all-cause 28-day in-hospital mortality rate of 29.9%, a median ICU stay of 8 days (interquartile range [IQR], 3 to 17 days), a median hospital stay of 13 days (IQR, 7 to 25 days), and an all-cause 30-day readmission rate (among nonhospice survivors) of 10.8%. Mortality decreased over time, from 43.5% (95% CI, 31.3% to 53.8%) to 19.2% (CI, 11.6% to 26.7%) between the first and last 15-day periods in the core adjusted model, whereas patient acuity and other factors did not change. LIMITATIONS: Single-health system study; use of, or highly dynamic trends in, other clinical interventions were not evaluated, nor were complications. CONCLUSION: Among patients with COVID-19-related critical illness admitted to ICUs of a learning health system in the United States, mortality seemed to decrease over time despite stable patient characteristics. Further studies are necessary to confirm this result and to investigate causal mechanisms. PRIMARY FUNDING SOURCE: Agency for Healthcare Research and Quality.

Locally Informed Simulation to Predict Hospital Capacity Needs During the COVID-19 Pandemic.

BACKGROUND: The coronavirus disease 2019 (COVID-19) pandemic challenges hospital leaders to make time-sensitive, critical decisions about clinical operations and resource allocations. OBJECTIVE: To estimate the timing of surges in clinical demand and the best- and worst-case scenarios of local COVID-19-induced strain on hospital capacity, and thus inform clinical operations and staffing demands and identify when hospital capacity would be saturated. DESIGN: Monte Carlo simulation instantiation of a susceptible, infected, removed (SIR) model with a 1-day cycle. SETTING: 3 hospitals in an academic health system. PATIENTS: All people living in the greater Philadelphia region. MEASUREMENTS: The COVID-19 Hospital Impact Model (CHIME) (http://penn-chime.phl.io) SIR model was used to estimate the time from 23 March 2020 until hospital capacity would probably be exceeded, and the intensity of the surge, including for intensive care unit (ICU) beds and ventilators. RESULTS: Using patients with COVID-19 alone, CHIME estimated that it would be 31 to 53 days before demand exceeds existing hospital capacity. In best- and worst-case scenarios of surges in the number of patients with COVID-19, the needed total capacity for hospital beds would reach 3131 to 12 650 across the 3 hospitals, including 338 to 1608 ICU beds and 118 to 599 ventilators. LIMITATIONS: Model parameters were taken directly or derived from published data across heterogeneous populations and practice environments and from the health system's historical data. CHIME does not incorporate more transition states to model infection severity, social networks to model transmission dynamics, or geographic information to account for spatial patterns of human interaction. CONCLUSION: Publicly available and designed for hospital operations leaders, this modeling tool can inform preparations for capacity strain during the early days of a pandemic. PRIMARY FUNDING SOURCE: University of Pennsylvania Health System and the Palliative and Advanced Illness Research Center.

A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice.

OBJECTIVES: Develop and implement a machine learning algorithm to predict severe sepsis and septic shock and evaluate the impact on clinical practice and patient outcomes. DESIGN: Retrospective cohort for algorithm derivation and validation, pre-post impact evaluation. SETTING: Tertiary teaching hospital system in Philadelphia, PA. PATIENTS: All non-ICU admissions; algorithm derivation July 2011 to June 2014 (n = 162,212); algorithm validation October to December 2015 (n = 10,448); silent versus alert comparison January 2016 to February 2017 (silent n = 22,280; alert n = 32,184). INTERVENTIONS: A random-forest classifier, derived and validated using electronic health record data, was deployed both silently and later with an alert to notify clinical teams of sepsis prediction. MEASUREMENT AND MAIN RESULT: Patients identified for training the algorithm were required to have International Classification of Diseases, 9th Edition codes for severe sepsis or septic shock and a positive blood culture during their hospital encounter with either a lactate greater than 2.2 mmol/L or a systolic blood pressure less than 90 mm Hg. The algorithm demonstrated a sensitivity of 26% and specificity of 98%, with a positive predictive value of 29% and positive likelihood ratio of 13. The alert resulted in a small statistically significant increase in lactate testing and IV fluid administration. There was no significant difference in mortality, discharge disposition, or transfer to ICU, although there was a reduction in time-to-ICU transfer. CONCLUSIONS: Our machine learning algorithm can predict, with low sensitivity but high specificity, the impending occurrence of severe sepsis and septic shock. Algorithm-generated predictive alerts modestly impacted clinical measures. Next steps include describing clinical perception of this tool and optimizing algorithm design and delivery.

The Impact of a Computerized Clinical Decision Support Tool on Inappropriate Clostridium difficile Testing.

OBJECTIVE To evaluate the effectiveness of a computerized clinical decision support intervention aimed at reducing inappropriate Clostridium difficile testing DESIGN Retrospective cohort study SETTING University of Pennsylvania Health System, comprised of 3 large tertiary-care hospitals PATIENTS All adult patients admitted over a 2-year period INTERVENTION Providers were required to use an order set integrated into a commercial electronic health record to order C. difficile toxin testing. The order set identified patients who had received laxatives within the previous 36 hours and displayed a message asking providers to consider stopping laxatives and reassessing in 24 hours prior to ordering C. difficile testing. Providers had the option to continue or discontinue laxatives and to proceed with or forgo testing. The primary endpoint was the change in inappropriate C. difficile testing, as measured by the number of patients who had C. difficile testing ordered while receiving laxatives. RESULTS Compared to the 1-year baseline period, the intervention resulted in a decrease in the proportion of inappropriate C. difficile testing (29.6% vs 27.3%; P=.02). The intervention was associated with an increase in the number of patients who had laxatives discontinued and did not undergo C. difficile testing (5.8% vs 46.4%; P<.01) and who had their laxatives discontinued and underwent testing (5.4% vs 35.2%; P<.01). We observed a nonsignificant increase in the proportion of patients with C. difficile related complications (5.0% vs 8.9%; P=.11). CONCLUSIONS A C. difficile order set was successful in decreasing inappropriate C. difficile testing and improving the timely discontinuation of laxatives. Infect Control Hosp Epidemiol 2017;38:1204-1208.

Association Between Index Hospitalization and Hospital Readmission in Sepsis Survivors.

OBJECTIVES: Hospital readmission is common after sepsis, yet the relationship between the index admission and readmission remains poorly understood. We sought to examine the relationship between infection during the index acute care hospitalization and readmission and to identify potentially modifiable factors during the index sepsis hospitalization associated with readmission. DESIGN: In a retrospective cohort study, we evaluated 444 sepsis survivors at risk of an unplanned hospital readmission in 2012. The primary outcome was 30-day unplanned hospital readmission. SETTING: Three hospitals within an academic healthcare system. SUBJECTS: Four hundred forty-four sepsis survivors. MEASUREMENTS AND MAIN RESULTS: Of 444 sepsis survivors, 23.4% (95% CI, 19.6-27.6%) experienced an unplanned 30-day readmission compared with 10.1% (95% CI, 9.6-10.7%) among 11,364 nonsepsis survivors over the same time period. The most common cause for readmission after sepsis was infection (69.2%, 72 of 104). Among infection-related readmissions, 51.4% were categorized as recurrent/unresolved. Patients with sepsis present on their index admission who also developed a hospital-acquired infection ("second hit") were nearly twice as likely to have an unplanned 30-day readmission compared with those who presented with sepsis at admission and did not develop a hospital-acquired infection or those who presented without infection and then developed hospital-acquired sepsis (38.6% vs 22.2% vs 20.0%, p = 0.04). Infection-related hospital readmissions, specifically, were more likely in patients with a "second hit" and patients receiving a longer duration of antibiotics. The use of total parenteral nutrition (p = 0.03), longer duration of antibiotics (p = 0.047), prior hospitalizations, and lower discharge hemoglobin (p = 0.04) were independently associated with hospital readmission. CONCLUSIONS: We confirmed that the majority of unplanned hospital readmissions after sepsis are due to an infection. We found that patients with sepsis at admission who developed a hospital-acquired infection, and those who received a longer duration of antibiotics, appear to be high-risk groups for unplanned, all-cause 30-day readmissions and infection-related 30-day readmissions.

Clinician Perception of the Effectiveness of an Automated Early Warning and Response System for Sepsis in an Academic Medical Center.

RATIONALE: We implemented an electronic early warning and response system (EWRS) to improve detection of and response to severe sepsis. Sustainability of such a system requires stakeholder acceptance. We hypothesized that clinicians receiving such alerts perceive them to be useful and effective. OBJECTIVES: To survey clinicians after EWRS notification about perceptions of the system. METHODS: For a 6-week study period 1 month after EWRS implementation in a large tertiary referral medical center, bedside clinicians, including providers (physicians, advanced practice providers) and registered nurses (RNs), were surveyed confidentially within 2 hours of an alert. MEASUREMENTS AND MAIN RESULTS: For the 247 alerts that triggered, 127 providers (51%) and 105 RNs (43%) completed the survey. Clinicians perceived most patients as stable before and after the alert. Approximately half (39% providers, 48% RNs) felt the alert provided new information, and about half (44% providers, 56% RNs) reported changes in management as a result of the alert, including closer monitoring and additional interventions. Over half (54% providers, 65% RNs) felt the alert was appropriately timed. Approximately one-third found the alert helpful (33% providers, 40% RNs) and fewer felt it improved patient care (24% providers, 35% RNs). CONCLUSIONS: A minority of responders perceived the EWRS to be useful, likely related to the perception that most patients identified were stable. However, management was altered half the time after an alert. These results suggest further improvements to the system are needed to enhance clinician perception of the system's utility.

Post-Acute Care Use and Hospital Readmission after Sepsis.

RATIONALE: The epidemiology of post-acute care use and hospital readmission after sepsis remains largely unknown. OBJECTIVES: To examine the rate of post-acute care use and hospital readmission after sepsis and to examine risk factors and outcomes for hospital readmissions after sepsis. METHODS: In an observational cohort study conducted in an academic health care system (2010-2012), we compared post-acute care use at discharge and hospital readmission after 3,620 sepsis hospitalizations with 108,958 nonsepsis hospitalizations. We used three validated, claims-based approaches to identify sepsis and severe sepsis. MEASUREMENTS AND MAIN RESULTS: Post-acute care use at discharge was more likely after sepsis, driven by skilled care facility placement (35.4% after sepsis vs. 15.8%; P < 0.001), with the highest rate observed after severe sepsis. Readmission rates at 7, 30, and 90 days were higher postsepsis (P < 0.001). Compared with nonsepsis hospitalizations (15.6% readmitted within 30 d), the increased readmission risk was present regardless of sepsis severity (27.3% after sepsis and 26.0-26.2% after severe sepsis). After controlling for presepsis characteristics, the readmission risk was found to be 1.51 times greater (95% CI, 1.38-1.66) than nonsepsis hospitalizations. Readmissions after sepsis were more likely to result in death or transition to hospice care (6.1% vs. 13.3% after sepsis; P < 0.001). Independent risk factors associated with 30-day readmissions after sepsis hospitalizations included age, malignancy diagnosis, hospitalizations in the year prior to the index hospitalization, nonelective index admission type, one or more procedures during the index hospitalization, and low hemoglobin and high red cell distribution width at discharge. CONCLUSIONS: Post-acute care use and hospital readmissions were common after sepsis. The increased readmission risk after sepsis was observed regardless of sepsis severity and was associated with adverse readmission outcomes.

Development, implementation, and impact of an automated early warning and response system for sepsis.

BACKGROUND: Early recognition and timely intervention significantly reduce sepsis-related mortality. OBJECTIVE: Describe the development, implementation, and impact of an early warning and response system (EWRS) for sepsis. DESIGN: After tool derivation and validation, a preimplementation/postimplementation study with multivariable adjustment measured impact. SETTING: Urban academic healthcare system. PATIENTS: Adult non-ICU patients admitted to acute inpatient units from October 1, 2011 to October 31, 2011 for tool derivation, June 6, 2012 to July 5, 2012 for tool validation, and June 6, 2012 to September 4, 2012 and June 6, 2013 to September 4, 2013 for the preimplementation/postimplementation analysis. INTERVENTION: An EWRS in our electronic health record monitored laboratory values and vital signs in real time. If a patient had ≥4 predefined abnormalities at any single time, the provider, nurse, and rapid response coordinator were notified and performed an immediate bedside patient evaluation. MEASUREMENTS: Screen positive rates, test characteristics, predictive values, and likelihood ratios; system utilization; and resulting changes in processes and outcomes. RESULTS: The tool's screen positive, sensitivity, specificity, and positive and negative predictive values and likelihood ratios for our composite of intensive care unit (ICU) transfer, rapid response team call, or death in the derivation cohort was 6%, 16%, 97%, 26%, 94%, 5.3, and 0.9, respectively. Validation values were similar. The EWRS resulted in a statistically significant increase in early sepsis care, ICU transfer, and sepsis documentation, and decreased sepsis mortality and increased discharge to home, although neither of these latter 2 findings reached statistical significance. CONCLUSIONS: An automated prediction tool identified at-risk patients and prompted a bedside evaluation resulting in more timely sepsis care, improved documentation, and a suggestion of reduced mortality.

Usability and impact of a computerized clinical decision support intervention designed to reduce urinary catheter utilization and catheter-associated urinary tract infections.

OBJECTIVE: To evaluate the usability and effectiveness of a computerized clinical decision support (CDS) intervention aimed at reducing the duration of urinary tract catheterizations. DESIGN: Retrospective cohort study. SETTING: Academic healthcare system. PATIENTS: All adult patients admitted from March 2009 through May 2012. INTERVENTION: A CDS intervention was integrated into a commercial electronic health record. Providers were prompted at order entry to specify the indication for urinary catheter insertion. On the basis of the indication chosen, providers were alerted to reassess the need for the urinary catheter if it was not removed within the recommended time. Three time periods were examined: baseline, after implementation of the first intervention (stock reminder), and after a second iteration (homegrown reminder). The primary endpoint was the usability of the intervention as measured by the proportion of reminders through which providers submitted a remove urinary catheter order. Secondary endpoints were the urinary catheter utilization ratio and the rate of hospital-acquired catheter-associated urinary tract infections (CAUTIs). RESULT: The first intervention displayed limited usability, with 2% of reminders resulting in a remove order. Usability improved to 15% with the revised reminder. The catheter utilization ratio declined over the 3 time periods (0.22, 0.20, and 0.19, respectively; P < .001), as did CAUTIs per 1,000 patient-days (0.84, 0.70, and 0.51, respectively; P < .001). CONCLUSIONS: A urinary catheter removal reminder system was successfully integrated within a healthcare system's electronic health record. The usability of the reminder was highly dependent on its user interface, with a homegrown version of the reminder resulting in higher impact than a stock reminder.

The readmission risk flag: using the electronic health record to automatically identify patients at risk for 30-day readmission.

BACKGROUND: Identification of patients at high risk for readmission is a crucial step toward improving care and reducing readmissions. The adoption of electronic health records (EHR) may prove important to strategies designed to risk stratify patients and introduce targeted interventions. OBJECTIVE: To develop and implement an automated prediction model integrated into our health system's EHR that identifies on admission patients at high risk for readmission within 30 days of discharge. DESIGN: Retrospective and prospective cohort. SETTING: Healthcare system consisting of 3 hospitals. PATIENTS: All adult patients admitted from August 2009 to September 2012. INTERVENTIONS: An automated readmission risk flag integrated into the EHR. MEASURES: Thirty-day all-cause and 7-day unplanned healthcare system readmissions. RESULTS: Using retrospective data, a single risk factor, ≥ 2 inpatient admissions in the past 12 months, was found to have the best balance of sensitivity (40%), positive predictive value (31%), and proportion of patients flagged (18%), with a C statistic of 0.62. Sensitivity (39%), positive predictive value (30%), proportion of patients flagged (18%), and C statistic (0.61) during the 12-month period after implementation of the risk flag were similar. There was no evidence for an effect of the intervention on 30-day all-cause and 7-day unplanned readmission rates in the 12-month period after implementation. CONCLUSIONS: An automated prediction model was effectively integrated into an existing EHR and identified patients on admission who were at risk for readmission within 30 days of discharge.

Effectiveness of a novel and scalable clinical decision support intervention to improve venous thromboembolism prophylaxis: a quasi-experimental study.

BACKGROUND: Venous thromboembolism (VTE) causes morbidity and mortality in hospitalized patients, and regulators and payors are encouraging the use of systems to prevent them. Here, we examine the effect of a computerized clinical decision support (CDS) intervention implemented across a multi-hospital academic health system on VTE prophylaxis and events. METHODS: The study included 223,062 inpatients admitted between April 2007 and May 2010, and used administrative and clinical data. The intervention was integrated into a commercial electronic health record (EHR) in an admission orderset used for all admissions. Three time periods were examined: baseline (period 1), and the time after implementation of the first CDS intervention (period 2) and a second iteration (period 3). Providers were prompted to accept or decline prophylaxis based on patient risk. Time series analyses examined the impact of the intervention on VTE prophylaxis during time periods two and three compared to baseline, and a simple pre-post design examined impact on VTE events and bleeds secondary to anticoagulation. VTE prophylaxis and events were also examined in a prespecified surgical subset of our population meeting the public reporting criteria defined by the Agency for Healthcare Research and Quality (AHRQ) Patient Safety Indicator (PSI). RESULTS: Unadjusted analyses suggested that "recommended", "any", and "pharmacologic" prophylaxis increased from baseline to the last study period (27.1% to 51.9%, 56.7% to 78.1%, and 42.0% to 54.4% respectively; p < 0.01 for all comparisons). Results were significant across all hospitals and the health system overall. Interrupted time series analyses suggested that our intervention increased the use of "recommended" and "any" prophylaxis by 7.9% and 9.6% respectively from baseline to time period 2 (p < 0.01 for both comparisons); and 6.6% and 9.6% respectively from baseline to the combined time periods 2 and 3 (p < 0.01 for both comparisons). There were no significant changes in "pharmacologic" prophylaxis in the adjusted model. The overall percent of patients with VTE increased from baseline to the last study period (2.0% to 2.2%; p = 0.03), but an analysis excluding patients with VTE "present on admission" (POA) demonstrated no difference in events (1.3% to 1.3%; p = 0.80). Overall bleeds did not significantly change. An analysis examining VTE prophylaxis and events in a surgical subset of patients defined by the AHRQ PSI demonstrated increased "recommended", "any", and "pharmacologic" prophylaxis from baseline to the last study period (32.3% to 60.0%, 62.8% to 85.7%, and 47.9% to 63.3% respectively; p < 0.01 for all comparisons) as well as reduced VTE events (2.2% to 1.7%; p < 0.01). CONCLUSIONS: The CDS intervention was associated with an increase in "recommended" and "any" VTE prophylaxis across the multi-hospital academic health system. The intervention was also associated with increased VTE rates in the overall study population, but a subanalysis using only admissions with appropriate POA documentation suggested no change in VTE rates, and a prespecified analysis of a surgical subset of our sample as defined by the AHRQ PSI for public reporting purposes suggested reduced VTE. This intervention was created in a commonly used commercial EHR and is scalable across institutions with similar systems.

Correlates of osteoprotegerin and association with aortic pulse wave velocity in patients with chronic kidney disease.

BACKGROUND AND OBJECTIVES: Osteoprotegerin (OPG), a cytokine that regulates bone resorption, has been implicated in the process of vascular calcification and stiffness. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: Serum OPG was measured in 351 participants with chronic kidney disease (CKD) from one site of the Chronic Renal Insufficiency Cohort Study. Cortical bone mineral content (BMC) was measured by quantitative computed tomography in the tibia. Multivariable linear regression was used to test the association between serum OPG and traditional cardiovascular risk factors, measures of abnormal bone and mineral metabolism, and pulse wave velocity. RESULTS: Higher serum OPG levels were associated with older age, female gender, greater systolic BP, lower estimated GFR, and lower serum albumin. OPG was not associated with measures of abnormal bone or mineral metabolism including serum phosphorus, albumin-corrected serum calcium, intact parathyroid hormone, bone-specific alkaline phosphatase, or cortical BMC. Among 226 participants with concurrent aortic pulse wave velocity measurements, increasing tertiles of serum OPG were associated with higher aortic pulse wave velocity after adjustment for demographics, traditional vascular risk factors, and nontraditional risk factors such as estimated GFR, albuminuria, serum phosphate, corrected serum calcium, presence of secondary hyperparathyroidism, serum albumin, and C-reactive protein or after additional adjustment for cortical BMC in a subset (n = 161). CONCLUSIONS: These data support a strong relationship between serum OPG and arterial stiffness independent of many potential confounders including traditional cardiovascular risk factors, abnormal bone and mineral metabolism, and inflammation.

Overall and cancer related mortality among patients with ocular inflammation treated with immunosuppressive drugs: retrospective cohort study.

CONTEXT: Whether immunosuppressive treatment adversely affects survival is unclear. OBJECTIVE: To assess whether immunosuppressive drugs increase mortality. DESIGN: Retrospective cohort study evaluating overall and cancer mortality in relation to immunosuppressive drug exposure among patients with ocular inflammatory diseases. Demographic, clinical, and treatment data derived from medical records, and mortality results from United States National Death Index linkage. The cohort's mortality risk was compared with US vital statistics using standardised mortality ratios. Overall and cancer mortality in relation to use or non-use of immunosuppressive drugs within the cohort was studied with survival analysis. SETTING: Five tertiary ocular inflammation clinics. Patients 7957 US residents with non-infectious ocular inflammation, 2340 of whom received immunosuppressive drugs during follow up. Exposures Use of antimetabolites, T cell inhibitors, alkylating agents, and tumour necrosis factor inhibitors. MAIN OUTCOME MEASURES: Overall mortality, cancer mortality. RESULTS: Over 66 802 person years (17 316 after exposure to immunosuppressive drugs), 936 patients died (1.4/100 person years), 230 (24.6%) from cancer. For patients unexposed to immunosuppressive treatment, risks of death overall (standardised mortality ratio 1.02, 95% confidence interval [CI] 0.94 to 1.11) and from cancer (1.10, 0.93 to 1.29) were similar to those of the US population. Patients who used azathioprine, methotrexate, mycophenolate mofetil, ciclosporin, systemic corticosteroids, or dapsone had overall and cancer mortality similar to that of patients who never took immunosuppressive drugs. In patients who used cyclophosphamide, overall mortality was not increased and cancer mortality was non-significantly increased. Tumour necrosis factor inhibitors were associated with increased overall (adjusted hazard ratio [HR] 1.99, 95% CI 1.00 to 3.98) and cancer mortality (adjusted HR 3.83, 1.13 to 13.01). CONCLUSIONS: Most commonly used immunosuppressive drugs do not seem to increase overall or cancer mortality. Our results suggesting that tumour necrosis factor inhibitors might increase mortality are less robust than the other findings; additional evidence is needed.