Danielle M. Nash

Detecting chronic kidney disease in population-based administrative databases using an algorithm of hospital encounter and physician claim codes

BackgroundLarge, population-based administrative healthcare databases can be used to identify patients with chronic kidney disease (CKD) when serum creatinine laboratory results are unavailable. We examined the validity of algorithms that used combined hospital encounter and physician claims database codes for the detection of CKD in Ontario, Canada.MethodsWe accrued 123,499 patients over the age of 65 from 2007 to 2010. All patients had a baseline serum creatinine value to estimate glomerular filtration rate (eGFR). We developed an algorithm of physician claims and hospital encounter codes to search administrative databases for the presence of CKD. We determined the sensitivity, specificity, positive and negative predictive values of this algorithm to detect our primary threshold of CKD, an eGFR <45 mL/min per 1.73 m2 (15.4% of patients). We also assessed serum creatinine and eGFR values in patients with and without CKD codes (algorithm positive and negative, respectively).ResultsOur algorithm required evidence of at least one of eleven CKD codes and 7.7% of patients were algorithm positive. The sensitivity was 32.7% [95% confidence interval: (95% CI): 32.0 to 33.3%]. Sensitivity was lower in women compared to men (25.7 vs. 43.7%; p <0.001) and in the oldest age category (over 80 vs. 66 to 80; 28.4 vs. 37.6 %; p < 0.001). All specificities were over 94%. The positive and negative predictive values were 65.4% (95% CI: 64.4 to 66.3%) and 88.8% (95% CI: 88.6 to 89.0%), respectively. In algorithm positive patients, the median [interquartile range (IQR)] baseline serum creatinine value was 135 μmol/L (106 to 179 μmol/L) compared to 82 μmol/L (69 to 98 μmol/L) for algorithm negative patients. Corresponding eGFR values were 38 mL/min per 1.73 m2 (26 to 51 mL/min per 1.73 m2) vs. 69 mL/min per 1.73 m2 (56 to 82 mL/min per 1.73 m2), respectively.ConclusionsPatients with CKD as identified by our database algorithm had distinctly higher baseline serum creatinine values and lower eGFR values than those without such codes. However, because of limited sensitivity, the prevalence of CKD was underestimated.

Atypical Antipsychotic Drugs and the Risk for Acute Kidney Injury and Other Adverse Outcomes in Older Adults: A Population-Based Cohort Study

Context Acute kidney injury (AKI) is reportedly associated with atypical antipsychotic drugs, although the risk has not been quantified. Contribution This population-based cohort study found that persons who had received a prescription for any of 3 atypical antipsychotic drugs in the previous 90 days had an elevated risk for hospitalization with AKI. These drugs were also associated with increased risk for hypotension, acute urinary retention, and death. Caution Only older adults and 3 antipsychotic agents were studied. Implication An association with specific adverse events may explain the increased risk for AKI observed with certain atypical antipsychotic drugs. The Editors Each year, millions of older adults worldwide are prescribed atypical antipsychotic drugs (quetiapine, risperidone, and olanzapine). These drugs are frequently used to manage behavioral symptoms of dementia, which is not an approved indication, and such use has raised safety concerns (1, 2). These drugs antagonize -adrenergic, muscarinic, serotonin, and dopamine receptors (3). Acute kidney injury (AKI) (defined as a sudden loss of kidney function) from atypical antipsychotic drugs is described in several case reports (48). Adverse outcomes potentially attributable to these drugs, such as hypotension, acute urinary retention, and the neuroleptic malignant syndrome or rhabdomyolysis, are known to cause AKI (411). Moreover, pneumonia, acute myocardial infarction, and ventricular arrhythmia have been associated with these drugs in previous population-based studies and AKI may also co-occur with these events (1214). However, no clinical or epidemiologic studies have quantified the risk for AKI from atypical antipsychotic drugs and information on outcomes of hypotension, acute urinary retention, and the neuroleptic malignant syndrome or rhabdomyolysis is limited. Such information would contribute to growing knowledge of potential adverse events from this drug class. The U.S. Food and Drug Administration warns of an increased risk for death in older patients treated with these drugs based on analyses of randomized, placebo-controlled trials (averaging 10 weeks in duration) (1). For these reasons, we did this population-based study of older adults to investigate the 90-day risk for hospitalization with AKI and other adverse outcomes from new use of an oral atypical antipsychotic drug initiated in the nonhospital setting. Methods Design and Setting We conducted this study at the Institute for Clinical Evaluative Sciences according to a prespecified protocol that was approved by the research ethics board at Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada. Patient informed consent was not required. We did a population-based, retrospective cohort study of older adults using linked health care databases in Ontario, Canada. Ontario residents have universal access to hospital care and physician services, and those aged 65 years or older have universal prescription drug coverage. The reporting of this study followed guidelines for observational studies (Table 1 of the Supplement) (15). Supplement. Supplementary material Data Sources We ascertained patient characteristics, drug use, covariate information, and outcome data using records from 5 databases. We obtained vital statistics from the Registered Persons Database of Ontario, which contains demographic information on all Ontario residents who have ever been issued a health card. We used the Ontario Drug Benefit database to identify prescription drug use. This database contains highly accurate recordsthe error rate is less than 1%of all outpatient prescriptions dispensed to patients aged 65 years or older (16). We identified diagnostic and procedural information on all hospitalizations from the Canadian Institute for Health Information Discharge Abstract Database. We obtained covariate information from the Ontario Health Insurance Plan database, which includes health claims for inpatient and outpatient physician services. We identified diagnostic information on all admissions to adult mental health beds from the Ontario Mental Health Reporting System. We have used these databases to research adverse drug reactions and health outcomes (including AKI) (1722). The databases were complete for all variables used in this study, except for prescriber information (which was missing for 10.8% of patients in the cohort). Codes from the International Classification of Diseases, Ninth Revision (before 2002), and Tenth Revision (after 2002), were used to assess baseline comorbid conditions in the 5 years before cohort entry (Table 2 of the Supplement). Codes used to ascertain outcomes are detailed in Table 3 of the Supplement, which lists only codes from the Tenth Revision because all events would have occurred after implementation of that coding system. A subpopulation in southwestern Ontario had information on outpatient serum creatinine levels available before cohort entry; this group was in the catchment area of 12 hospitals in which linked laboratory values were also available (23). Patients We established a cohort of older adults with evidence of a new outpatient prescription for an oral atypical antipsychotic drug (quetiapine, risperidone, or olanzapine) between June 2003 and December 2011. The date of this prescription served as the index date (cohort entry date) for the drug recipients. We matched a group of drug nonrecipients similar in health status to the recipients. We randomly assigned an index date to the entire Ontario population according to the index date of the drug recipients. For example, if more recipients had an index date between 2003 and 2005, a greater proportion of the population would have been randomly assigned an index date between 2003 and 2005. From these adults, after applying our exclusions to both groups, we matched a drug nonrecipient to each recipient on the following 11 characteristics: age (within 2 years); sex; residential status (community-dwelling or long-term care); evidence of comorbid conditions (dementia, schizophrenia or other psychotic disorder, bipolar disorder, major depression or anxiety disorder, Parkinson disease, and chronic kidney disease); constituency in the subpopulation with available information on serum creatinine levels; and the logit of the propensity score for the predicted probability of newly receiving an atypical antipsychotic drug (within a caliper of 0.2 SDs). We derived this propensity score from a logistic regression model and selected 91 variables for inclusion in the score on the basis of their potential association with the study outcomes or atypical antipsychotic drug initiation (variables listed in Table 4 of the Supplement) (24). One of the variables was the Johns Hopkins Adjusted Clinical Group Aggregated Diagnosis Groups (a validated measure of the complexity of comorbid conditions based on groups of diagnoses) (25, 26). Before matching, we excluded the following patients from both groups: those with prescriptions for any antipsychotic drug in the 180 days before their index date to ensure that the drug was newly prescribed (or had the potential to be newly prescribed in the case of the nonrecipients); those who were discharged from a hospital in the 2 days before their index date to ensure that drug use was newly initiated in the nonhospitalized setting (as in Ontario, patients continuing atypical antipsychotic drug treatment initiated in a hospital would have their oral outpatient prescription dispensed the day of or the day after hospital discharge); and those with evidence of end-stage renal disease before their index date (because the development of AKI is no longer relevant). Among the drug recipients, those who received a prescription for more than 1 type of antipsychotic drug (for example, a prescription for quetiapine and olanzapine) on their index date were excluded to compare mutually exclusive groups in subgroup analyses. Among the nonrecipients, those who did not have at least 1 outpatient medication dispensed in the 90 days before their index date were excluded to ensure that such persons were able to receive a prescription. Each drug recipient and nonrecipient could be selected only once for cohort entry. Outcomes We followed patients for 90 days after the index date to assess the prespecified outcomes. We chose 90 days to focus on acute adverse events, avoid potential crossovers between the 2 groups that might occur with longer follow-up, and mimic the duration of follow-up described in clinical trials of atypical antipsychotic drugs in older patients (1, 2, 27). The primary outcome was hospitalization with AKI. The secondary adverse outcomes were known causes of AKI (hospitalization with hypotension, acute urinary retention, the neuroleptic malignant syndrome or rhabdomyolysis, pneumonia, acute myocardial infarction, and ventricular arrhythmia) and all-cause mortality. The diagnosis codes used to identify the outcomes and information on their accuracy are presented in Table 3 of the Supplement (2830). For hospitalization records, up to 25 diagnosis codes can be assigned per hospitalization (for example, codes for AKI or rhabdomyolysis). Therefore, patients with codes for multiple study outcomes were accounted for in the assessment of each outcome. We previously examined the validity of the database code for hospitalization with AKI used in the current study. In this previous validation study (30), the database code for AKI identified a median increase in serum creatinine level of 98 mol/L (1.11 mg/dL) (interquartile range [IQR], 43 to 200 mol/L [0.49 to 2.26 mg/dL]) at the time of hospital presentation from the most recent value before hospitalization. The absence of such a code represented no statistically significant change in serum creatinine level (6 mol/L [0.07 mg/dL]; IQR, 4 to 20 mol/L [0.05 to 0.23 mg/dL]) (30). Although specificity was greater than 95%, the sensitivity of the hospital diagnosis was

Rehospitalizations and Emergency Department Visits after Hospital Discharge in Patients Receiving Maintenance Hemodialysis

Clinical outcomes after a hospital discharge are poorly defined for patients receiving maintenance in-center (outpatient) hemodialysis. To describe the proportion and characteristics of these patients who are rehospitalized, visit an emergency department, or die within 30 days after discharge from an acute hospitalization, we conducted a population-based study of all adult patients receiving maintenance in-center hemodialysis who were discharged between January 1, 2003, and December 31, 2011, from 157 acute care hospitals in Ontario, Canada. For patients with more than one hospitalization, we randomly selected a single hospitalization as the index hospitalization. Of the 11,177 patients included in the final cohort, 1926 (17%) were rehospitalized, 2971 (27%) were treated in the emergency department, and 840 (7.5%) died within 30 days of discharge. Complications of type 2 diabetes mellitus were the most common reason for rehospitalization, whereas heart failure was the most common reason for an emergency department visit. In multivariable analysis using a cause-specific Cox proportional hazards model, the following characteristics were associated with 30-day rehospitalization: older age, the number of hospital admissions in the preceding 6 months, the number of emergency department visits in the preceding 6 months, higher Charlson comorbidity index score, and the receipt of mechanical ventilation during the index hospitalization. Thus, a large proportion of patients receiving maintenance in-center hemodialysis will be readmitted or visit an emergency room within 30 days of an acute hospitalization. A focus on improving care transitions from the inpatient setting to the outpatient dialysis unit may improve outcomes and reduce healthcare costs.

Risk of kidney stones with surgical intervention in living kidney donors.

A kidney stone in a person with a solitary kidney requires urgent attention, which may result in surgical and/or hospital attention. We conducted a matched retrospective cohort study to determine if living kidney donors compared to healthy nondonors have a higher risk of: (i) kidney stones with surgical intervention, and (ii) hospital encounters for kidney stones. We reviewed all predonation charts for living kidney donations from 1992 to 2009 at five major transplant centers in Ontario, Canada, and linked this information to healthcare databases. We selected nondonors from the healthiest segment of the general population and matched 10 nondonors to every donor. Of the 2019 donors and 20 190 nondonors, none had evidence of kidney stones prior to cohort entry. Median follow‐up time was 8.4 years (maximum 19.7 years; loss to follow‐up <7%). There was no difference in the rate of kidney stones with surgical intervention in donors compared to nondonors (8.3 vs. 9.7 events/10 000 person‐years; rate ratio 0.85; 95% confidence interval [CI] 0.47–1.53). Similarly there was no difference in the rate of hospital encounters for kidney stones (12.1 vs. 16.1 events/10 000 person‐years; rate ratio 0.75; 95% CI 0.45–1.24). These interim results are reassuring for the safety of living kidney donation.

Macrolide antibiotics and the risk of ventricular arrhythmia in older adults

Background: Many respiratory tract infections are treated with macrolide antibiotics. Regulatory agencies warn that these antibiotics increase the risk of ventricular arrhythmia. We examined the 30-day risk of ventricular arrhythmia and all-cause mortality associated with macrolide antibiotics relative to nonmacrolide antibiotics. Methods: We conducted a population-based retrospective cohort study involving older adults (age > 65 yr) with a new prescription for an oral macrolide antibiotic (azithromycin, clarithromycin or erythromycin) in Ontario from 2002 to 2013. Our primary outcome was a hospital encounter with ventricular arrhythmia within 30 days after a new prescription. Our secondary outcome was 30-day all-cause mortality. We matched patients 1:1 using propensity scores to patients prescribed nonmacrolide antibiotics (amoxicillin, cefuroxime or levofloxacin). We used conditional logistic regression to measure the association between macrolide exposure and outcomes, and repeated the analysis in 4 subgroups defined by the presence or absence of chronic kidney disease, congestive heart failure, coronary artery disease and concurrent use of a drug known to prolong the QT interval. Results: Compared with nonmacrolide antibiotics, macrolide antibiotics were not associated with a higher risk of ventricular arrhythmia (0.03% v. 0.03%; relative risk [RR] 1.06, 95% confidence interval [CI] 0.83–1.36) and were associated with a lower risk of all-cause mortality (0.62% v. 0.76%; RR 0.82, 95% CI 0.78–0.86). These associations were similar in all subgroups. Interpretation: Among older adults, macrolide antibiotics were not associated with a higher 30-day risk of ventricular arrhythmia than nonmacrolide antibiotics. These findings suggest that current warnings from the US Food and Drug Administration may be overstated.

Risk of serious gastrointestinal bleeding in living kidney donors

Individuals with moderate‐to‐severe reduced renal function have greater risk of gastrointestinal bleeding than those with normal renal function. We conducted a retrospective matched cohort study to assess whether living kidney donors share a similar risk. We reviewed pre‐donation charts for living kidney donations from 1992 to 2009 in Ontario, Canada, and linked this information to healthcare databases. We selected healthy non‐donors from the general population and matched ten non‐donors to every donor. Of the 2009 donors and 20 090 matched non‐donors, none had evidence of gastrointestinal bleeding prior to cohort entry. The cohort was followed for a median of 8.4 yr (maximum 19.7 yr; loss to follow‐up <7%). There was no significant difference in the rate of hospitalization with gastrointestinal bleeding in donors compared to non‐donors (18.5 vs. 14.9 events per 10 000 person‐years; rate ratio 1.24; 95% confidence interval [CI] 0.85–1.81). Similar results were obtained when we assessed the time to first hospitalization with gastrointestinal bleeding (hazard ratio 1.25, 95% CI 0.87–1.79). In conclusion, we found living kidney donation was not associated with a higher risk of hospitalization with gastrointestinal bleeding. These results are reassuring for the safety of the practice.

30-Day Readmissions After an Acute Kidney Injury Hospitalization

BACKGROUND The risk of hospital readmission in acute kidney injury survivors is not well understood. We estimated the proportion of acute kidney injury patients who were rehospitalized within 30 days and identified characteristics associated with hospital readmission. METHODS We conducted a population-based study of patients who survived a hospitalization complicated by acute kidney injury from 2003-2013 in Ontario, Canada. The primary outcome was 30-day hospital readmission. We used a propensity score model to match patients with and without acute kidney injury, and a Cox proportional hazards model with death as a competing risk to identify predictors of 30-day readmission. RESULTS We identified 156,690 patients who were discharged from 197 hospitals after an episode of acute kidney injury. In the subsequent 30 days, 27,457 (18%) patients were readmitted; 15,988 (10%) visited the emergency department and 7480 (5%) died. We successfully matched 111,778 patients with acute kidney injury 1:1 to patients without acute kidney injury. The likelihood of 30-day readmission was higher in acute kidney injury patients than those without acute kidney injury (hazard ratio [HR] 1.53; 95% confidence interval [CI], 1.50-1.57). Factors most strongly associated with 30-day rehospitalization were the number of hospitalizations in the preceding year (adjusted HR 1.45 for ≥2 hospitalizations; 95% CI, 1.40-1.51) and receipt of inpatient chemotherapy (adjusted HR 1.44; 95% CI, 1.32-1.58). CONCLUSIONS One in 5 patients who survive a hospitalization complicated by acute kidney injury is readmitted in the next 30 days. Better strategies are needed to identify and care for acute kidney injury survivors in the community.

Comparing two types of macrolide antibiotics for the purpose of assessing population-based drug interactions

Objective Clarithromycin strongly inhibits enzyme cytochrome P450 3A4, preventing the metabolism of some other drugs, while azithromycin is a weak inhibitor. Accordingly, blood concentrations of other drugs increase with clarithromycin coprescription leading to adverse events. These macrolide antibiotics also differ on other properties that may impact outcomes. In this study, we compared outcomes in two groups of macrolide antibiotic users in the absence of potentially interacting drugs. Design Population-based retrospective cohort study. Setting Ontario, Canada, from 2003 to 2010. Patients Patients (mean 74 years) prescribed clarithromycin (n=52 251) or azithromycin (referent group, n=46 618). Main outcomes The primary outcomes were hospital admission within 30 days of a new antibiotic prescription with any of the 12 conditions examined separately (acute kidney injury, acute myocardial infarction, neuroimaging (proxy for delirium), hypotension, syncope, hyperkalaemia, hyponatraemia, hyperglycaemia, arrhythmia, ischaemic stroke, gastrointestinal bleeding and sepsis). The secondary outcome was mortality. Results The baseline characteristics of the two groups, including patient demographics, comorbid conditions, infection type and prescribing physician specialty, were nearly identical. The median daily dose was 1000 mg for clarithromycin and 300 mg for azithromycin and the median duration of dispensing antibiotics was 10 and 5 days, respectively. There was no difference between the groups in the risk of hospitalisation for any condition studied (relative risk ranged from 0.67 to 1.23). Compared with azithromycin, clarithromycin was associated with a slightly higher risk of all-cause mortality (0.46% vs 0.37%, relative risk 1.25, 95% CI 1.03 to 1.52). Conclusions Clarithromycin can be used to assess drug interactions in population-based studies with azithromycin serving as a control group. However, any differences in mortality observed between the two antibiotic groups in the setting of other drug use may be partially attributable to factors beyond the inhibition of drug metabolising enzymes and transporters, as the difference for this outcome was significant.

Characteristics and Outcomes of Patients Discharged Home from an Emergency Department with AKI

BACKGROUND AND OBJECTIVES Patients discharged home from an emergency department with AKI are not well described. This study describes their characteristics and outcomes and compares these outcomes to two referent groups. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS We conducted a population-based retrospective cohort study in Ontario, Canada from 2003 to 2012 of 6346 patients aged ≥40 years who were discharged from the emergency department with AKI (defined using serum creatinine values). We analyzed the risk of all-cause mortality, receipt of acute dialysis, and hospitalization within 30 days after discharge. We used propensity score methods to compare all-cause mortality to two referent groups. We matched 4379 discharged patients to 4379 patients who were hospitalized from the emergency department with similar AKI stage. We also matched 6188 discharged patients to 6188 patients who were discharged home from the emergency department with no AKI. RESULTS There were 6346 emergency department discharges with AKI. The mean age was 69 years and 6012 (95%) had stage 1, 290 (5%) had stage 2, and 44 (0.7%) had stage 3 AKI. Within 30 days, 149 (2%) (AKI stage 1: 127 [2%]; stage 2: 15 [5%]; stage 3: seven [16%]) died, 22 (0.3%) received acute dialysis, and 1032 (16%) were hospitalized. An emergency department discharge versus hospitalization with AKI was associated with lower mortality (3% versus 12%; relative risk, 0.3; 95% confidence interval, 0.2 to 0.3). An emergency department discharge with AKI versus no AKI was associated with higher mortality (2% versus 1%; relative risk, 1.6; 95% confidence interval, 1.2 to 2.0). CONCLUSIONS Patients discharged home from the emergency department with AKI are at risk of poor 30-day outcomes. A better understanding of care in this at-risk population is warranted, as are testing strategies to improve care.

Patterns and Predictors of Screening for Breast and Cervical Cancer in Women with CKD

BACKGROUND AND OBJECTIVES Breast and cervical cancers are prevalent in women with CKD, but it is uncertain how often screening for these cancers should be undertaken given concerns that the benefits of screening may be fewer and the harms greater in women with CKD than in the general population. We examined patterns of breast and cervical cancer screening in women on the basis of CKD stage and age and assessed predictors of screening. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS We conducted two population-based cohort studies (breast and cervical cancer screening) from 2002 to 2013 using linked administrative health care data from Ontario, Canada. A total of 141,326 and 324,548 women were included in the breast and cervical cancer screening cohorts, respectively. RESULTS The 2-year cumulative incidences were 61% among women without CKD, 54% for those with CKD stages 3a and 3b, 37% for those with CKD stages 4 and 5, and 26% for women on dialysis. Similar patterns were observed for the 3-year cumulative incidences of cervical cancer screening. The associations of breast and cervical cancer screening with CKD were modified by age and CKD stage, where lower incidence of screening in women with advanced CKD compared with no CKD was most pronounced in older age groups (P<0.001). Older age, higher comorbidity burden, and lower-income groups were associated with a lower rate of screening. CONCLUSIONS Most women with advanced CKD do not receive breast or cervical cancer screening. A better understanding of patient and health professional preferences toward cancer screening in CKD is needed along with the outcomes of such screening.