Are risk evaluation and mitigation strategies associated with less off-label use of medications? The case of immune thrombocytopenia.

Using data from a large commercial health insurer, we studied prescribing of romiplostim (Nplate) and eltrombopag (Promacta), two drugs for primary immune thrombocytopenia (ITP) for which risk evaluation and mitigation strategies (REMS) with elements to assure safe use were initially imposed and then removed. We identified 103 and 117 new users of romiplostim and eltrombopag, respectively. Use was almost exclusively for FDA-approved indications ("on-label") while the REMS with elements to assure safe use were in place. After these elements were lifted, off-label use of eltrombopag among patients with hepatitis C virus (HCV), a subsequently approved indication, increased. The ratio of incidence rate ratios of off-label/HCV to on-label initiation of eltrombopag between the two time periods was significant (13.41; P < 0.001). Our finding of an association with reduced off-label prescribing suggests that REMS with elements to assure safe use can help promote patient safety but may also prevent promising off-label drug uses.

Clinical and regulatory features of drugs not initially approved by the FDA.

US Food and Drug Administration (FDA) denials of New Drug Applications (NDAs) are often controversial, yet little is known about the basis and characteristics of these decisions. We reviewed new drugs and biologics evaluated by FDA advisory committees between 2007 and 2009 and found that half (27/52, 52%) were unapproved in the first cycle. By 2013, 63% had been approved. Products with initial safety concerns (62%) were much more likely to be approved eventually than drugs with efficacy concerns (8%).

Defining "innovativeness" in drug development: a systematic review.

Some observers of drug development argue that the pace of pharmaceutical innovation is declining, but others deny that contention. This controversy may be due to different methods of defining and assessing innovation. We conducted a systematic review of the literature to develop a taxonomy of methods for measuring innovation in drug development. The 42 studies fell into four main categories: counts of new drugs approved, assessments of therapeutic value, economic outcomes, and patents issued. The definition determined whether a study found a positive or negative trend in innovative drug development. Of 21 studies that relied on counts, 9 (43%) concluded that the trend for drug discovery was favorable, 11 (52%) concluded that the trend was not favorable, and 1 reached no conclusion. By contrast, of 21 studies that used other measures of innovation, 0 concluded that the trend was favorable, 8 (47%) concluded that the trend was not favorable, and 13 reached no conclusion (P = 0.03).

Defining the epidemiology of bisphosphonate-associated osteonecrosis of the jaw: prior work and current challenges.

UNLABELLED: Bisphosphonate-related osteonecrosis of the jaw (BONJ) is an adverse effect of bisphosphonate use with a poorly described epidemiology in osteoporosis patients. We examined the literature and two new cohorts for BONJ. The literature suggests an incidence rate of 0.028 % to 4.3 %. Our cohort studies found an incidence of 0.02 % (95 % CI 0.004 %-0.11 %). INTRODUCTION: We examined the epidemiology of BONJ associated with osteoporosis dosing of bisphosphonates. METHODS: First, we systematically searched the literature about osteoporosis BONJ. Identified studies were abstracted by two authors. Second, we attempted to estimate the relative risk of BONJ among bisphosphonate users with osteoporosis. Two different large insurance databases, one from 2005-2007 and another from 2007-2010, combined with medical record review, were searched. The older dataset did not include the International Classification of Diagnoses (ICD) diagnosis code for osteonecrosis of the jaw (ONJ; ICD 733.45). Incidence rates and relative risks were estimated using Cox regression. RESULTS: The literature review produced nine studies of varying quality. The incidence rates for BONJ among osteoporosis patients varied from 0.028 % to 4.3 %. Two prior studies estimated the relative risk of ONJ related to bisphosphonates and found odds ratios of 7.2 and 9.2. Our attempts to estimate the incidence rate of BONJ encompassed 41,957 in the dataset from 2005-2007 and 466,645 in a separate dataset from 2007-2010. From the older dataset, we found 51 potential cases of BONJ using a broad definition of possible ONJ. One case was confirmed by a dentist for a prevalence of 0.02 % (95 % CI 0.004 %-0.11 %) among bisphosphonate users. From the newer dataset, we found 13 possible cases, but none could be confirmed. Most subjects with the ONJ diagnosis code appeared to have had an osteoporosis-related fracture and not ONJ. CONCLUSIONS: The literature suggests a broad range of possible values for the prevalence of BONJ; our estimate fell within the range from prior literature.

Differential risk of death in older residents in nursing homes prescribed specific antipsychotic drugs: population based cohort study.

OBJECTIVE: To assess risks of mortality associated with use of individual antipsychotic drugs in elderly residents in nursing homes. DESIGN: Population based cohort study with linked data from Medicaid, Medicare, the Minimum Data Set, the National Death Index, and a national assessment of nursing home quality. SETTING: Nursing homes in the United States. PARTICIPANTS: 75,445 new users of antipsychotic drugs (haloperidol, aripiprazole, olanzapine, quetiapine, risperidone, ziprasidone). All participants were aged ≥ 65, were eligible for Medicaid, and lived in a nursing home in 2001-5. MAIN OUTCOME MEASURES: Cox proportional hazards models were used to compare 180 day risks of all cause and cause specific mortality by individual drug, with propensity score adjustment to control for potential confounders. RESULTS: Compared with risperidone, users of haloperidol had an increased risk of mortality (hazard ratio 2.07, 95% confidence interval 1.89 to 2.26) and users of quetiapine a decreased risk (0.81, 0.75 to 0.88). The effects were strongest shortly after the start of treatment, remained after adjustment for dose, and were seen for all causes of death examined. No clinically meaningful differences were observed for the other drugs. There was no evidence that the effect measure modification in those with dementia or behavioural disturbances. There was a dose-response relation for all drugs except quetiapine. CONCLUSIONS: Though these findings cannot prove causality, and we cannot rule out the possibility of residual confounding, they provide more evidence of the risk of using these drugs in older patients, reinforcing the concept that they should not be used in the absence of clear need. The data suggest that the risk of mortality with these drugs is generally increased with higher doses and seems to be highest for haloperidol and least for quetiapine.

Refilling and switching of antiepileptic drugs and seizure-related events.

We sought to estimate the risk of seizure-related events associated with refilling prescriptions for antiepileptic drugs (AEDs) and to estimate the effect of switching between brand-name and generic drugs or between two generic versions of the same drug. We conducted a case-crossover study using health-care databases from British Columbia, Canada, among AED users who had an emergency room visit or hospitalization for seizure (index seizure-related event), defined using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) codes 345.xx (epilepsy and recurrent seizures) and 780.3x (convulsions), between 1997 and 2005. AED prescription refilling itself was associated with 2.3-fold elevated odds of seizure-related events when the refill occurred within 21 days before the index event (odds ratio (OR) 2.31; 95% confidence interval (CI) 1.56-3.44). The OR was 2.75 (95% CI 0.88-8.64) for refills that involved switching, yielding a refill-adjusted OR for switching of 1.19 (95% CI 0.35-3.99). Refilling the same AED prescription was associated with an elevated risk of seizure-related events whether or not the refill involved switching from a brand-name to a generic product.

A blinded randomized controlled trial of motivational interviewing to improve adherence with osteoporosis medications: design of the OPTIMA trial.

UNLABELLED: We have designed an innovative randomized controlled trial for improving adherence with osteoporosis medications. Recruitment and randomization have been successful. Also, the counseling intervention has been well accepted by subjects randomized to this treatment arm. INTRODUCTION: While many effective treatments exist for osteoporosis, most people do not adhere to such treatments long term. No proven interventions exist to improve osteoporosis medication adherence. We report here on the design and initial enrollment in an innovative randomized controlled trial aimed at improving adherence to osteoporosis treatments. METHODS: The trial represents a collaboration between academic researchers and a state-run pharmacy benefits program for low-income older adults. Beneficiaries beginning treatment with a medication for osteoporosis are targeted for recruitment. We randomize consenting individuals to receive 12 months of mailed education (control arm) or an intervention consisting of one-on-one telephone-based counseling and the mailed education. Motivational interviewing forms the basis for the counseling program which is delivered by seven trained and supervised health counselors over ten telephone calls. The counseling sessions include scripted dialog and open-ended questions about medication adherence and its barriers, as well as structured questions. The primary end point of the trial is medication adherence measured over the 12-month intervention period. Secondary end points include fractures, nursing home admissions, health care resource utilization, and mortality. RESULTS: During the first 7 months of recruitment, we have screened 3,638 potentially eligible subjects. After an initial mailing, 1,115 (30.6%) opted out of telephone recruitment and 1,019 (28.0%) could not be successfully contacted. Of the remaining, 879 (24.2%) consented to participate and were randomized. Women comprise over 90% of all groups; mean ages range from 77 to 80 years old, and the majority in all groups was white. The distribution of osteoporosis medications was comparable across groups and the median number of different prescription drugs used in the prior year was eight to ten. CONCLUSIONS: We have developed a novel intervention for improving osteoporosis medication adherence. The intervention is currently being tested in a large-scale randomized controlled trial. If successful, the intervention may represent a useful model for improving adherence to other chronic treatments.

Interventions to improve adherence and persistence with osteoporosis medications: a systematic literature review.

SUMMARY: Adherence and persistence with osteoporosis medications are poor. We conducted a systematic literature review of interventions to improve adherence and persistence with osteoporosis medications. Seven studies met eligibility requirements and were included in the review. Few interventions were efficacious, and no clear trends regarding successful intervention techniques were identified. However, periodic follow-up interaction between patients and health professionals appeared to be beneficial. INTRODUCTION: Adherence and persistence with pharmacologic therapy for osteoporosis are suboptimal. Our goal was to examine the design and efficacy of published interventions to improve adherence and persistence. METHODS: We searched medical literature databases for English-language papers published between January 1990 and July 2008. We selected papers that described interventions and provided results for control and intervention subjects. We assessed the design and methods of each study, including randomization, blinding, and reporting of drop-outs. We summarized the results and calculated effect sizes for each trial. RESULTS: Seven studies met eligibility requirements and were included in the review. Five of the seven studies provided adherence data. Of those five studies, three showed a statistically significant (p < or = 0.05) improvement in adherence by the intervention group, with effect sizes from 0.17 to 0.58. Five of the seven studies provided persistence data. Of those five, one reported statistically significant improvement in persistence by the intervention group, with an effect size of 0.36. CONCLUSIONS: Few interventions were efficacious, and no clear trends regarding successful intervention techniques were identified in this small sample of studies. However, periodic follow-up interaction between patients and health professionals appeared to be beneficial.

Uric acid lowering therapy: prescribing patterns in a large cohort of older adults.

BACKGROUND: Uric acid lowering therapy (UALT) is considered a chronic treatment for gout. Relatively little is known about adherence to UALT. METHODS: We assessed adherence with UALT over a 1-year study period among 9823 older adults enrolled in a pharmacy benefit program. Two adherence measures were calculated, the percentage of days covered (PDC) and the time until an extended break (at least 60 days) in treatment. A PDC <80% was considered poor adherence and its predictors were examined in multivariable logistic models. RESULTS: The mean (SD) PDC was 54% (36%) with 64% of patients considered poorly compliant over the study period. A total of 56% had experienced an extended break in UALT. Predictors of poor adherence included younger age (odds ratio (OR) 1.50, 95% CI 1.33-1.69 for ages 65-74 compared with 85 and above) and African-American race (OR 1.86, 95% CI 1.52-2.27 compared with Caucasian race). Most patients (93%) received their initial UALT prescription from a non-specialist and this also predicted poor adherence (OR 1.15, 95% CI 0.96-1.38 compared with rheumatologists or nephrologists). CONCLUSION: Adherence with UALT is poor. While uric acid levels were not measured in this study, poor adherence with UALT is likely to reduce attainment of goal uric acid levels.

Clinical presentation of myocardial infarction contributes to lower use of coronary angiography in patients with chronic kidney disease.

Patients with chronic kidney disease (CKD) have high mortality following myocardial infarction (MI), but are less likely to undergo coronary angiography than those without CKD. Whether this phenomenon is explained by differences in the presentation of MI or by bias against performing coronary angiography in patients with CKD is unclear. We examined the clinical presentation of 1876 elderly patients who presented with MI and categorized them by estimated glomerular filtration rate: >60 ml/min (no/mild CKD), 30-60 ml/min (CKD Stage 3) or <30 ml/min (CKD Stage 4/5). Compared with patients with no/mild CKD, patients with CKD Stage 3 or Stage 4/5 had more comorbidity, greater prior nursing home use, and higher frequency of conduction abnormalities or anterior infarction. By contrast, peak creatinine kinase-MB fraction (CK-MB) concentrations were lower and ST-elevation MI was less common in patients with CKD Stage 3 or Stage 4/5. In univariate analyses, patients with CKD Stage 4/5 (odds ratio (OR)=0.34, 95% confidence interval (CI): 0.23-0.50) or Stage 3 (OR=0.57, 95% CI: 0.45-0.73) were markedly less likely to undergo angiography than subjects with no/mild CKD. After multivariable adjustment, the association of CKD Stage 3 with the use of coronary angiography was attenuated (OR=0.78, 95% CI: 0.60-1.03), but CKD Stage 4/5 remained strongly associated with lower use (OR=0.52, 95% CI: 0.34-0.80). Clinical features of MI are different in patients with and without CKD and may partly explain the low use of angiography in patients with CKD Stage 3. However, the clinical features of MI do not account for its underuse in MI patients with CKD Stages 4/5. Whether reduced use of angiography in patients with advanced CKD is justified must be evaluated in formal risk-benefit analyses.

Predialysis nephrologist care and access to kidney transplantation in the United States.

Predialysis nephrologist care is associated with morbidity and mortality in incident dialysis patients, but the relationship with access to kidney transplantation (KT) is unclear. From a national study of incident US dialysis patients, we identified 2253 patients with detailed information about predialysis care, sociodemographic characteristics and comorbidities. We used multivariate Cox proportional hazards models to study associations between predialysis nephrology care and two outcomes: time from first dialysis to the first day on the KT wait-list, and time to first KT. Two-thirds of patients first encountered a nephrologist >3 months prior to dialysis and one-third

Patterns of cardiovascular risk in rheumatoid arthritis.

BACKGROUND: Although it is known that rheumatoid arthritis is associated with an increased risk of cardiovascular disease (CVD), the pattern of this risk is not clear. This study investigated the relative risk of myocardial infarction, stroke and CVD mortality in adults with rheumatoid arthritis compared with adults without rheumatoid arthritis across age groups, sex and prior CVD event status. METHODS: We conducted a cohort study among all residents aged >or=18 years residing in British Columbia between 1999 and 2003. Residents who had visited the doctor at least thrice for rheumatoid arthritis (International Classification of Disease = 714) were considered to have rheumatoid arthritis. A non-rheumatoid arthritis cohort was matched to the rheumatoid arthritis cohort by age, sex and start of follow-up. The primary composite end point was a hospital admission for myocardial infarction, stroke or CVD mortality. RESULTS: 25 385 adults who had at least three diagnoses for rheumatoid arthritis during the study period were identified. During the 5-year study period, 375 patients with rheumatoid arthritis had a hospital admission for myocardial infarction, 363 had a hospitalisation for stroke, 437 died from cardiovascular causes and 1042 had one of these outcomes. The rate ratio for a CVD event in patients with rheumatoid arthritis was 1.6 (95% confidence interval (CI) 1.5 to 1.7), and the rate difference was 5.7 (95% CI 4.9 to 6.4) per 1000 person-years. The rate ratio decreased with age, from 3.3 in patients aged 18-39 years to 1.6 in those aged >or=75 years. However, the rate difference was 1.2 per 1000 person-years in the youngest age group and increased to 19.7 per 1000 person-years in those aged >or=75 years. Among patients with a prior CVD event, the rate ratios and rate differences were not increased in rheumatoid arthritis. CONCLUSIONS: This study confirms that rheumatoid arthritis is a risk factor for CVD events and shows that the rate ratio for CVD events among subjects with rheumatoid arthritis is highest in young adults and those without known prior CVD events. However, in absolute terms, the difference in event rates is highest in older adults.

A randomized controlled trial of mailed osteoporosis education to older adults.

INTRODUCTION: Osteoporosis represents a growing public health concern; however, current rates of management are sub-optimal. The aim of our study was to assess, in a randomized controlled trial, the effect of a mailed educational intervention on older adults' knowledge, attitudes, and preventive behaviors regarding osteoporosis. The setting was a large publicly funded state pharmacy benefits program. The patients were 31,715 Medicare beneficiaries from Pennsylvania who participated in a drug benefits program for low-to-moderate income elderly people. METHODS: All women aged over 65 years, and all men and women with a history of fracture or long-term oral use of glucocorticoid, were included. Approximately half of the participants (intervention group) were randomly selected to receive three mailings aimed at improving knowledge of osteoporosis and enhancing preventive activities, such as using calcium and vitamin D, reducing fall risks in the home, obtaining a bone mineral density (BMD) test, and taking medications when necessary. The other participants did not receive the intervention mailings and served as controls. We surveyed a sample of intervention and control subjects to determine the effects of the intervention on knowledge, attitudes, self-efficacy (confidence in one's ability to perform specific activities), and behavior regarding osteoporosis prevention and treatment. Six hundred randomly selected participants in the intervention group and an equal number in the control group were invited to participate. RESULTS: Twenty-six had died and 636 of the remaining 1,185 (54%) completed the survey. Respondents and non-respondents did not differ significantly with respect to measured sociodemographic factors. All scales had good reliability (all Cronbach's alphas>0.65). Knowledge of osteoporosis was generally very good and did not differ between intervention (mean=65% correct responses) and control subjects (mean=67% correct; P=0.4). Perceived susceptibility to osteoporosis was relatively high and similar across groups (P=0.4). Self-efficacy for participating in osteoporosis prevention and treatment was very strong in both the intervention (mean=4.3 on a 0-5 scale) and control (mean=4.2, P=0.03) groups . On average, subjects in the intervention group reported participating in 3.5 of 6 preventive osteoporosis activities compared with 3.4 in the control group (P=0.5). CONCLUSIONS: Compared with the controls, a mailed educational intervention for osteoporosis was not associated with better knowledge, higher perceived susceptibility, or performance of preventive measures among the at-risk older adults that we studied. The intervention group demonstrated a small increase in self-efficacy. More intensive patient interventions or intervention aimed at other aspects of the care process may be required to bring about changes that lead to a reduction in fractures.

A cost minimization model for the treatment of minor bleeding episodes in patients with haemophilia A and high-titre inhibitors.

Treatment of acute bleeding episodes in patients with haemophilia A and inhibitory antibodies to factor VIII (FVIII) most often involves the use of bypassing haemostatic agents, such as activated prothrombin complex concentrates (aPCC) or recombinant factor VIIa (rFVIIa). We constructed a cost minimization model to compare the costs of initial treatment with aPCC vs. rFVIIa in the home treatment of minor bleeding episodes. We developed a clinical scenario describing such a case and presented it to a panel of US haemophilia specialists. For each product class, we asked panellists to provide dosing regimens required to achieve complete resolution of a minor haemarthrosis in a child with high-titre inhibitors, and for the probabilities of success at two time points (8-12 and 24 h). Consensus among the panellists was refined by a second round of the process, and the median values resulting were used as inputs to a decision analysis model. Sensitivity analyses were conducted to determine threshold values for key variables. The base case model found that initial treatment with aPCC would result in a mean cost per episode of 21 000 dollars, compared with 33 400 dollars for initial treatment with rFVIIa. Sensitivity analyses over a range of clinically plausible values for cost, dosing, and efficacy did not change the selection of aPCC as the dominant strategy.

The economic impact of factor VIII inhibitors in patients with haemophilia.

The impact on the cost of care for haemophilia patients with inhibitors is not well defined. To quantify the effect on health care expenditures associated with inhibitors to factor VIII (FVIII) or FIX, we conducted a retrospective cohort study examining product use and outcomes in adult and paediatric haemophilia patients with and without inhibitors. Twelve patients with inhibitors to FVIII or FIX (cases) identified in the haemophilia surveillance system (HSS) at two centres were matched on age, severity of haemophilia, and treatment centre to haemophilia patients without inhibitors. Patients with HIV or significant liver disease were excluded from the study. All eligible non-inhibitor control patients were selected for inclusion in the study, resulting in a total of 28 controls. We then tracked product usage and hospitalizations from programme entry until 1998 or loss to follow-up, producing a total database of 184 person-years of experience. A descriptive matched analysis was conducted to examine annual differences in the cost of product used and hospitalizations. We found that the median cost for factor products among haemophilia patients with inhibitors was $55,853/year, $2,760 less than comparable haemophilia patients without inhibitors. The median number of hospitalizations per year was 1.0 for both inhibitor and non-inhibitor patients and the median number of days hospitalized was virtually the same. Although these findings do not appear to support the belief that there is a substantial increase in the cost of care for haemophilia patients with inhibitors, it does document that a few outlier patients can drive the cost of treatment for this disease. As the largest component of the cost of care is that of factor concentrate, it becomes imperative in the current health care environment to better define the true costs and benefits of treatments designed to eradicate or manage inhibitors. A careful cost accounting of immune tolerance induction (ITI) and other therapeutic strategies, taking into account successes and failures and duration and intensity of therapy, should help to better define the costs and benefits of such approaches. Methods to identify high cost inhibitor patients should be developed so that these strategies may be targeted to appropriate candidates.

Undertreatment of hypercholesterolaemia: a population-based study.

AIMS: To assess the level of undertreatment of hypercholesterolaemia in the general population, taking intra-person variability in serum cholesterol concentrations into account, and to identify determinants of undertreatment of hypercholesterolaemia. METHODS: In this cross-sectional study, data from two population-based surveys on cardiovascular disease risk factors conducted between 1987 and 1997 in the Netherlands were used. For all 64 757 respondents aged 20-59 years, treatment eligibility for lipid-lowering drug use was established according to the Dutch Cholesterol Consensus. Multivariate logistic models were used to identify determinants of undertreatment. RESULTS: During the study period, 56.8% of the study population had undesirable cholesterol concentrations (serum total cholesterol> 5 mmol l(-1)) and 5.5% of those were eligible for pharmacological treatment based on their absolute risk of coronary heart disease. Of those eligible for pharmacological treatment, 16.3% were treated, and 19.6% of those treated had their serum total cholesterol concentration controlled. Only 3.2% of those eligible for pharmacological treatment were both treated and controlled. We identified several determinants for undertreatment, e.g. male gender and younger age for primary prevention and female gender and older age for secondary prevention. Treatment has improved slightly in more recent years. CONCLUSIONS: Over 95% of the population eligible for the pharmacological treatment of hypercholesterolaemia was either untreated or was uncontrolled. To decrease undertreatment, identification of high-risk patients should be increased. Those who are treated with lipid-lowering medication could further benefit from more aggressive treatment, especially with statins.

Porcine factor VIII: pharmacoeconomics of inhibitor therapy.

The treatment of acquired haemophilia is characteristically exceedingly expensive and thus a cost-benefit analysis of the several available treatment strategies is urgently needed. To address this issue, decision-analysis techniques were used to construct a cost-minimization model to compare the cost of treatment with porcine factor VIII (pFVIII), human FVIII (hFVIII) or an activated prothrombin complex concentrate (APCC). This model was based upon the results of a comprehensive literature search of all relevant clinical studies and case series. To supplement these data, a panel of haemophilia specialists was presented with a clinical scenario describing an acquired haemophilia patient with an acute haemorrhage in whom the human and porcine inhibitor titres were initially unknown. Based on this scenario and on their own clinical experience, the expert panel assessed the applicability of the model as initially constructed, assigned probabilities of success to each treatment and recommended appropriate initial dosing and follow-up regimens. This information was incorporated into the model and a simulation was conducted from which the costs of care were calculated. Sensitivity analyses were then conducted on all parameters. The results of the model show that treatment initiated with pFVIII would be more cost effective compared with treatment sequences initiated with an APCC or hFVIII, respectively. The model indicates that initial treatment with pFVIII in this scenario may be the preferred strategy clinically, as well as on economic grounds.

Determinants of delayed nephrologist referral in patients with chronic kidney disease.

Late referral to nephrologists of patients with chronic kidney disease (CKD) is a major public health problem because it is prevalent and associated with increased morbidity, mortality, and greater healthcare costs. To identify factors associated with delayed nephrologist referral (first nephrologist visit < 90 days before the onset of renal replacement therapy), we identified a cohort of patients with preexisting CKD that progressed to end-stage renal failure. We developed a logistic regression model to measure the association of specific demographic and clinical covariates with delayed nephrologist referral. Delayed referral was highly associated with older age (P < 0.001), race other than white or black (P = 0.002), and the absence of certain comorbidities: hypertension (P < 0.001), coronary artery disease (P < 0.001), malignancy (P = 0.005), and diabetes (P = 0.02). Associations of late referral with male sex (P = 0.07) and lower socioeconomic status (P = 0.09) were of borderline significance. Patients who were predominantly cared for by a general internist were more likely to be referred late to a nephrologist compared with those cared for by a family or primary care practitioner (P = 0.002) or another subspecialist (P = 0.019). These findings suggest that several factors increase the risk that patients with CKD will have the first nephrologist consultation excessively late in the course of their disease. Although timely access to nephrologist services is important for all patients with advanced CKD, this is of particular concern in older patients, those in certain minority populations, and those in whom the absence of comorbidity may provide a false sense of true risk status.