High Comorbid Condition Burden Associated with Withdrawal from Dialysis Therapy

Over the past decade, annual rates of dialysis therapy initiation have been relatively stable. Dialysis patients are living longer than before, creating an interest in issues such as conservative care for end-stage renal disease (ESRD), palliative care in nephrology, and withdrawal from dialysis therapy. Withdrawal from dialysis therapy is a cause of death in ~10% to 20% of patients in western countries and seems to be increasing. There are few data available on differences in clinical outcomes between patients on maintenance dialysis who do and do not withdraw from dialysis therapy.

James B. Wetmore, MD, MS, and colleagues conducted a case-control analysis to examine clinical events in the period preceding withdrawal in patients on maintenance dialysis. The study examined rates of medical events, time spent in the hospital and in skilled nursing facilities, and putative markers of morbidity; data were drawn in part from claims for durable medical equipment use in the period preceding withdrawal to assess differences in patients who withdraw and those who do not withdraw. The researchers sought to test the hypothesis that the period before withdrawal would be characterized by increasing rates of medical events, institutionalization, and other markers potentially signaling morbidity. Results of the study were reported in the American Journal of Kidney Diseases [2018;71(6):831-841].

The study utilized the US Renal Data System ESRD database that consists of data from the ESRD Medical Evidence Report, the ESRD Death Notification form, and Medicare Parts A and B claims. Medicare insures the majority of patients on maintenance dialysis; Medicare billing claims data were used to determine the presence of comorbid conditions, derive the Liu comorbidity index, and generate a putative marker of morbidity based on claims for durable medical equipment use.

Case patients were patients who withdrew from hemodialysis therapy between January 1, 2008, and December 31, 2011; for those patients, the researchers created an index date, defined as the date of withdraw, for each patient who withdrew. For hemodialysis patients who did not withdraw, the researchers created their respective index dates, defined as the calendar date on which dialysis duration was within ±30 days of dialysis therapy duration among patients who withdrew.

Medical events included in the analysis were hospitalizations for myocardial infarction (MI), congestive heart failure, stroke, amputation/critical limb ischemia, sepsis, pneumonia, vascular access infection, gastrointestinal bleeding, or fracture. The subset of hospitalizations for MI, stroke, amputation/critical limb ischemia, sepsis, or fractures were deemed “major medical events.”

There were 36,814 patients who initiated maintenance hemodialysis on or before January 1, 2011, and withdrew January 1, 2008, through December 31, 2011. Of those, 18,412 survived at least 1 year on hemodialysis therapy, and had at least 9 months of Medicare Parts A and B coverage immediately preceding their respective index dates. Of those, 18,367 were adults of known race for whom comparable nonwithdrawal patients could be found on the basis of the index date. The 18,367 patients were the case patients for the match 1 analysis and the matches (n=220,443) were the controls.

Of the patients in the withdrawal group, 13,322 were randomly further matched in a 1:4 ratio with nonwithdrawers with similar age, sex, race, cause of ESRD, and duration of dialysis therapy as of the index date, resulting in 53,288 more fully matched nonwithdrawers (match 2 group).

Among patients in the match 1 group, those who withdrew were older (mean age, 75.3 vs 66.2 years) and more likely to be female, be white than of a minority race/ethnicity (71.8% vs 46.2%), and have longer duration of dialysis therapy (4.8 vs 4.5 years). In addition, patients who withdrew had higher comorbid conditions and morbidity scores. In the match 2 group, there were only slight differences in distributions of matched variables; however, patients who withdrew had generally higher comorbid condition burdens and higher comorbidity scores.

In the match 1 group, there was an association between older age and withdraw from dialysis. Compared with patients 65 to 74 years of age, adjusted odds ratios (ORs) for withdrawal were 1.61 (95% confidence interval [ CI], 1.54-1.68) for patients 75 to 84 years of age and 2.68 (95% CI, 2.54-2.82) for those ≥85 years of age; the adjusted OR for patients 18 to 44 years of age was 0.36 (95% CI, 0.32-0.40).

Women were more likely to withdraw than men (adjusted OR, 1.07; 95% CI, 1.04-1.11). With the exception of Native Americans, compared with white patients, nonwhite patients had lower odds of withdrawal: adjusted ORs 0.36 (95% CI, 0.35-0.38) for blacks. 0.47 (95% CI, 0.42-0.53) for Asians, and 0.46 (95% CI, 0.44-0.49) for Hispanics. There was an association between longer duration of dialysis therapy and withdrawal, e.g., compared with duration of 3 to <4 years, adjusted ORs for withdrawal were 0.55 (95% CI, 0.52-0.58) for 1 to <2 years but 1.37 (95% CI, 1.30-1.44) for ≥5 years.

There was an association between hospital or skilled nursing facility stay and increased odds of withdrawal (1.01 per each day of institutionalization). Higher morbidity score was associated with increased likelihood of withdrawal: compared with a score of 0, a score of 3 to 4 was associated with an adjusted OR of 3.48 (95% CI, 3.29-3.67), and a score of ≥7, with an adjusted OR of 12.10 (95% CI, 11.37-12.87).

Findings were similar for the match 2 group for institutionalization, comorbid conditions, and morbidity score.

In the months preceding withdrawal, the rates of medical events and institutionalization tended to increase, as did morbidity score.

Limitations cited by the authors included the inability to generalize the findings to populations other than US Medicare beneficiaries, and not including data on patients who withdrew <1 year following initiation of dialysis therapy.

“In conclusion, the period before dialysis therapy withdrawal was characterized by increasing rates of medical events and hospitalization. A marker of morbidity based on durable medical equipment use appears to be strongly associated with withdrawal, even after adjustment for a host of other factors. Men and members of certain minority groups, such as blacks, Asians, and Hispanics, were less likely to withdraw than women and whites. When they did withdraw, blacks and Hispanics were more likely than whites to withdraw in the hospital. Further work examining elective dialysis therapy withdrawal experience is warranted,” the researchers said.

Takeaway Points

  1. Researchers conducted a case-control analysis to examine differences in the clinical course between patients receiving maintenance dialysis who do and who do not withdraw from dialysis therapy.
  2. Older age (≥85 years) was associated with higher adjusted odds of withdrawal and patients who withdrew were more likely to be female and of white race.
  3. Patients with higher comorbid condition burden were more likely to withdraw from dialysis. There was an association between a higher durable medical equipment claims-based morbidity score and the odds of withdraw; the association remained following adjustment for traditional comorbid scores and hospitalization.