Article

Clinical Outcomes after Parathyroidectomy in a Nationwide Cohort of Patients on Hemodialysis

| Areef Ishani,*†‡ Jiannong Liu,* James B. Wetmore,* Kimberly A. Lowe,§ Thy Do,§ Brian D. Bradbury,§ Geoffrey A. Block, and Allan J. Collins*

Abstract Background and objectives Patients receiving undergo parathyroidectomy to improve laboratory parameters in resistant with the assumption that clinical outcomes will also improve. *Chronic Disease fi Research Group, However, no randomized clinical trial data demonstrate the bene ts of parathyroidectomy. This study aimed to Minneapolis Medical evaluate clinical outcomes up to 1 year after parathyroidectomy in a nationwide sample of patients receiving Research Foundation, hemodialysis. Minneapolis, Minnesota; †Minneapolis Veterans Design, setting, participants, & measurements Using data from the US Renal Data System, this study identified $ Affairs Health Care prevalent hemodialysis patients aged 18 years with Medicare as primary payers who underwent parathy- System, Minneapolis, roidectomy from 2007 to 2009. Baseline characteristics and comorbid conditions were assessed in the year Minnesota; ‡ preceding parathyroidectomy; clinical events were identified in the year preceding and the year after Department of parathyroidectomy. After parathyroidectomy, patients were censored at death, loss of Medicare coverage, Medicine, University fi of Minnesota, kidney transplant, change in dialysis modality, or 365 days. This study estimated cause-speci ceventratesfor Minneapolis, both periods and rate ratios comparing event rates in the postparathyroidectomy versus preparathyroidectomy Minnesota; §Center for periods. Observational Research, Amgen Inc, Thousand Oaks, Results Of 4435 patients who underwent parathyroidectomy, 2.0% died during the parathyroidectomy California; and | hospitalization and the 30 days after discharge. During the 30 days after discharge, 23.8% of patients were Denver Nephrology rehospitalized; 29.3% of these patients required intensive care. In the year after parathyroidectomy, hospital- Clinical Research izations were higher by 39%, hospital days by 58%, intensive care unit admissions by 69%, and emergency room/ Division, Denver, observation visits requiring treatment by 20-fold compared with the preceding year. Cause- Colorado specific hospitalizations were higher for acute myocardial infarction (rate ratio, 1.98; 95% confidence interval, Correspondence: 1.60 to 2.46) and dysrhythmia (rate ratio 1.4; 95% confidence interval1.16 to 1.78); fracture rates did not differ (rate fi Dr. Areef Ishani, ratio 0.82; 95% con dence interval 0.6 to 1.1). Chronic Disease Research Group, Conclusions Parathyroidectomy is associated with significant morbidity in the 30 days after hospital discharge Minneapolis Medical and in the year after the procedure. Awareness of clinical events will assist in developing evidence-based Research Foundation, fi 914 South 8th Street, risk/bene t determinations for the indication for parathyroidectomy. Suite S4.100, Clin J Am Soc Nephrol 10: 90–97, 2015. doi: 10.2215/CJN.03520414 Minneapolis, MN 55404. Email: [email protected]

Introduction KDIGO guidelines currently state that patients with Secondary hyperparathyroidism (SHPT) is common CKD stages 3–5D with severe hyperparathyroidism among patients receiving dialysis (1). A major focus who fail to respond to medical therapy should un- of therapy for these patients is directed at control- dergo parathyroidectomy. This recommendation ling (PTH) levels, because may have been based on observational, not-randomized large, population-based observational studies have clinical trial evidence describing potentially benefi- suggested an association between severely elevated cial effects on short-term laboratory parameters and/or PTH values and poor patient outcomes (2,3). In re- on longer-term clinical outcomes including fracture and cent years, therapy for the management of SHPT has death (5,6). Some (7–12) but not all (5) reports suggest undergone several changes based on recommenda- that short-term adverse outcomes related to para- tions from the Kidney Disease Outcomes Quality produce only modest adverse conse- Initiative (4) and the Kidney Disease Improving quences. However, many of these reports are based Global Outcomes (KDIGO) international guideline on single-center experiences, and may represent group (3). Unfortunately, the guideline working highly selected and high-performing surgical units groups have had little randomized clinical trial (7–12). In addition, the comparability of nonpara- data available to inform high-quality recommendations thyroidectomized control groups in these studies is for care. unknown.

90 Copyright © 2015 by the American Society of Nephrology www.cjasn.org Vol 10 January, 2015 Clin J Am Soc Nephrol 10: 90–97, January, 2015 Clinical Outcomes after Parathyroidectomy, Ishani et al. 91

There is a paucity of data describing the clinical clinical outcomes of interest were all-cause mortality outcomes that occur within the year after parathyroid- and hospitalization events including total number of hos- ectomy. Using data from the US Renal Data System pitalizations, specific causes, intensive care unit (ICU) (USRDS), we sought to (1) evaluate morbidity and stays, total number of hospital days, outpatient visits in- mortality after a parathyroidectomy procedure and (2) cluding emergency department visits, hospital obser- compare event rates in the year immediately after para- vation stays, and outpatient physician visits with thyroidectomy with rates in the year immediately preced- hypocalcemia. We also ascertained the total number of ing it, using a nationwide sample of United States dialysis fractures. The process for identifying fractures, which patients who underwent parathyroidectomy between used a combination of ICD-9-CM and Current Procedural 2007 and 2009. Terminology codes and place of service, and a complete listing of the codes themselves are detailed in Supple- mental Appendix 1. For hospitalizations and outpatient Materials and Methods visits, we determined whether hypocalcemia requiring Data Source, Populations, and Case Definition treatment was associated with the event. Treatment for This study used USRDS ESRD data, which include data hypocalcemia was defined by ICD-9-CM diagnosis code from the ESRD Medical Evidence Report (US Centers for 275.41, 275.49, or 275.5 with drug code J0610 or J0620. Medicare and Medicaid Services [CMS] form CMS-2728), Hospitalization with hypocalcemia was defined as any fi the ESRD Death Noti cation (form CMS-2746), the United hospitalization with ICD-9-CM diagnosis code 275.41, Network for Organ Sharing kidney transplant database, 275.49, or 275.5. Cause-specific hospitalizations were de- Medicare Part A institutional claims (inpatient, outpatient, fined using the following primary diagnosis codes: skilled nursing facility, home health, and hospice), and CVA/TIA 430–438; congestive heart failure 398.91; acute Medicare Part B physician claims (inpatient, outpatient, myocardial infarction 402.x1, 422.xx, 425.xx, 428.xx, 404.x1, and supplier). 408.x3, v421, or 410.xx; and dysrhythmia 426–427, v450, The study population included all prevalent patients or v533. aged $18 years receiving hemodialysis who underwent parathyroidectomy between January 1, 2007, and December 31, 2009. Patients were required to have Medicare as the Statistical Analyses The characteristics of hemodialysis patients undergoing primary insurance payer for both Part A and Part B and to parathyroidectomy were evaluated using descriptive have been receiving hemodialysis for at least 1 year before statistics means and standard deviations or medians and undergoing parathyroidectomy. Parathyroidectomy was 25th/75th percentiles for continuous variables, and counts identified from Medicare inpatient claims using Inter- and percentages for categorical variables. The study is a national Classification of Diseases, Ninth Revision, Clini- pre-post comparison of patients who underwent para- cal Modification (ICD-9-CM) procedure codes 06.8x and thyroidectomy, in which events occurring in the 1 year 06.95(13).Thedateoftheparathyroidectomywasconsid- after parathyroidectomy were compared with the same ered the index date, and the hospitalization during which events occurring in the 1 year before parathyroidectomy. parathyroidectomy occurred was considered the index The hospitalization for the parathyroidectomy procedure hospitalization. was not included as an outcome in either the prepara- thyroidectomy or the postparathyroidectomy timeframe, Patient Characteristics and Outcomes but mortality during that hospitalization was included. Patient characteristics were derived from the ESRD Med- Event rates were calculated for each outcome as the ical Evidence Report and Medicare claims. Characteristics number of events divided by follow-up time years. Rate assessed as of the index date included age, race (white, black, ratios (RRs) and 95% confidence intervals (95% CIs) were other), sex, primary ESRD cause (diabetes, hypertension, used to compare the occurrence of clinical events in the glomerulonephritis, other), body mass index, dialysis duration, 1-year periods after versus immediately preceding para- geographic region (18 US Renal Networks), and comorbid thyroidectomy. We present results for the population conditions (congestive heart failure, cerebrovascular accident/ overall and within strata of important baseline character- transient ischemic attack [CVA/TIA], atherosclerotic heart istics including demographics, dialysis characteristics, disease, peripheral vascular disease, dysrhythmia, other and comorbid conditions. Differences of RRs among strata cardiovascular diseases, and diabetes). Comorbid conditions were tested in the Poisson model with interactions between fi were de ned by the presence of ICD-9-CM claims (typically outcomes in the postparathyroidectomy period, the pre- two outpatient claims or one inpatient claim) in the year parathyroidectomy period, and the corresponding pa- preceding parathyroidectomy. These are methods previously tient characteristic. Because this was a pre-post used by the USRDS (14). comparison, the correlation can be artificially increased. The generalized estimating equation method was used Characterization of Outcomes to manage this possible correlation. All analyses were The postparathyroidectomy assessment period time- conducted using SAS software (version 9.2; SAS, Cary, frame was broken into three distinct periods: (1)theindex NC). hospitalization associated with the parathyroidectomy procedure, (2) the 30-day postdischarge period (to assess acute morbidity and mortality associated with the para- Results thyroidectomy procedure), and (3) the 1-year post- We identified a total of 7707 patients who under- discharge period (to assess longer-term outcomes). The went parathyroidectomy between 2007 and 2009. After 92 Clinical Journal of the American Society of Nephrology

exclusion criteria were applied, 4435 patients were avail- to 1.60) and ICU use by 69% (64.0 versus 37.9 per 100 able for analysis (Figure 1). Table 1 describes the baseline patient-years; RR, 1.69; 95% CI, 1.59 to 1.80). Finally, hos- characteristics of the final cohort. Most patients (88.4%) pitalizations with hypocalcemia were higher by 17-fold were aged ,65 years; 57.5% were black and 66.0% had (37.7 versus 2.2 per 100 patient-years; RR, 17.1; 95% CI, been on dialysis for longer than 5 years. 13.9 to 20.9; Figure 2B). The attributable risk associated with parathyroidectomy is described in Supplemental Figure 1. Short-Term Outcomes Among the cause-specific hospitalizations evaluated, Immediately after parathyroidectomy, 41 patients (0.9%) we found a higher incidence of hospitalizations for died during the index hospitalization and another 48 (1.1%) CVA/TIA (RR, 1.83; 95% CI, 1.38 to 2.43), acute died within 30 days after discharge, an overall procedure- myocardial infarction (RR, 1.98; 95% CI, 1.60 to 2.46), related mortality of 2.0%. and dysrhythmia (RR, 1.44; 95% CI, 1.16 to 1.78). There The median duration of the index hospitalization for the was a trend toward lower incidence of hospitalizations parathyroidectomy procedure was 4 days (25th/75th per- for congestive heart failure, which did not achieve centile, 3–7 days), and 24.5% of patients required ICU ad- statistical significance (RR, 0.90; 95% CI, 0.80 to 1.02) mission. Within 30 days of discharge, 23.8% of patients (Table 2). were rehospitalized, and 29.3% of these required an ICU The morbidity associated with parathyroidectomy was stay (Table 2). not treated exclusively in the inpatient setting. Patients who underwent parathyroidectomy experienced more One-Year Outcomes total emergency room or hospital observation visits One-year mortality among patients undergoing para- (378.3 versus 314.3 per 100 patient-years; RR, 1.2; 95% thyroidectomy was 9.8% (rate 10.99 per 100 patient- CI, 1.17 to 1.23; Figure 2C) and more emergency room or years). A total of 7571 hospitalizations occurred among hospital observation stays with hypocalcemia treatment 2832 unique individuals in the year after parathyroid- (48.8 versus 2.4 per 100 patient-years; RR, 20.4; 95% CI, ectomy, an average of 2.7 hospitalizations per person 16.8 to 24.9; Figure 2D). Total outpatient visits with hy- among those hospitalized (Table 2). Compared with the pocalcemia requiring treatment were also higher (11.9 ver- preceding year, all-cause hospitalizations were higher sus 0.7 per 100 patient-years; RR, 17.0; 95% CI, 11.8 to 24.5). by 39% (192.0 versus 138.6 per 100 patient-years; RR, We also evaluated fracture events. Fractures occurred in- 1.39; 95% CI, 1.34 to 1.44; Figure 2A). Overall, the total frequently before parathyroidectomy (n=95) and in the number of hospital days was higher by 58% (1245.0 ver- year after parathyroidectomy (n=69; RR, 0.82; 95% CI, sus 789.3 per 100 patient-years; RR, 1.58; 95% CI, 1.56 0.60 to 1.12).

Figure 1. | Patient flow chart. Clin J Am Soc Nephrol 10: 90–97, January, 2015 Clinical Outcomes after Parathyroidectomy, Ishani et al. 93

versus partial); results did not change materially and are not Table 1. Baseline characteristics of parathyroidectomy shown. patients

Characteristic All Patients, n (%) Discussion Total 4435 (100.0) In this nationwide study of patients undergoing surgical Age (yr) parathyroidectomy, we observed significant clinical out- 19–44 1764 (39.8) comes in the first 30 days and up through 1 year after the – 45 64 2154 (48.6) procedure. Most pronounced were high rehospitaliza- 65–74 410 (9.2) $ tion rates, more ICU visits, more hospital days, and 2% 75 107 (2.4) mortality within 30 days of hospital discharge after the Race procedure. We specifically evaluated both all-cause events White 1685 (38.0) and cause-specificevents(i.e., hospitalization with hypo- Black 2551 (57.5) fi Other 199 (4.5) calcemia), because cause-speci c outcomes are likely spe- Sex cific but lack sensitivity. The relative increase in adverse Men 2298 (51.8) outcomes at 1 year varied significantly by selected patient Women 2137 (48.2) characteristics. Of note, the absence of most comorbid Primary cause of ESRD conditions was associated with a greater relative increase Diabetes 1013 (22.8) in adverse outcomes. We hypothesize that this is due to Hypertension 1462 (33.0) the already high event rates in patients with comorbid Glomerulonephritis 934 (21.1) conditions in the year preceding parathyroidectomy, Other/unknown/missing 1026 (23.1) such that a similar absolute increase in events postpara- BMI (kg/m2) ,18 151 (3.4) thyroidectomy results in a smaller relative increase. 18 to ,25 1217 (27.4) These data provide important new evidence regarding 25 to ,30 1026 (23.1) adverse clinical outcomes related to surgical parathyroid- 30 to ,35 772 (17.4) ectomy, which should be recognized when considering 35 to ,40 517 (11.7) this procedure as a treatment option for patients with $40 536 (12.1) severe SHPT. Missing 216 (4.9) The treatment of uncontrolled PTH elevations remains Dialysis duration (yr) , controversial and the best approach to therapy is unknown. 1to 3 538 (12.1) Given the uncertainty regarding the benefits and risks of 3to, 5 970 (21.9) parathyroidectomy, use of the procedure in the dialysis $52927(66.0) Comorbid conditions (%)a population has varied substantially over the past few Congestive heart failure 1973 (44.5) decades, ranging from a high of 12.5 per 1000 patient- CVA/TIA 522 (11.8) years in 1992 to a low of 5.5 per 1000 patient-years in 2005; ASHD 1542 (34.8) more recently, the rate was 9 per 1000 patient-years in Peripheral vascular disease 1389 (31.3) 2007 (15,16). Similar rates have been observed in other Dysrhythmia 1057 (23.8) countries (17). Cardiac disease, other 1623 (36.6) Based in part on observational studies consistently Diabetes 1954 (44.1) showing higher risk of adverse clinical outcomes in patients whose PTH levels are outside of the range be- BMI, body mass index; CVA/TIA; cerebrovascular accident/ tween 2–9 times the upper limit of normal for dialysis pa- transient ischemic attack; ASHD, atherosclerotic heart disease. tients, KDIGO recommends targeting PTH levels to within aComorbid conditions were identified from the Medical Evi- this range (3). However, few studies have comprehensively dence Report and medical claims during 12-month period be- fore parathyroidectomy. evaluated the risks related to parathyroidectomy. Studies that have examined the issue are reports from potentially high-performing, single surgical centers that have focused primarily on the benefits of parathyroidectomy. Almost uni- Subgroup Analyses formly, these studies show an improvement in laboratory We performed subgroup analysis by selected patient parameters. characteristics (Figure 3). Several subgroups evaluated A small number of studies have evaluated parathyroid- showed significant interactions with pre-post change in ectomy using a nationwide cohort of hemodialysis pa- risk of the outcomes of interest (P,0.05). Among the var- tients. Kestenbaum et al. (5) evaluated the effects of ious subgroup evaluated, those that appeared to experi- parathyroidectomy on survival using USRDS data ence greater change in risk of adverse outcomes included (1988–1999) and found that 30-day mortality was elevated increasing age (hospitalization, emergency department/ for patients who underwent parathyroidectomy com- observation stays), white and other race (hospitalization, pared with those who did not (3.1% versus 1.2%); this emergency department/observation visits), male sex (hos- elevated risk continued to 90 days and then inverted at pitalization, ICU admissions), a varying association with 1 year. The 30-day and 1-year mortality they report are body mass index (all outcomes), and absence of most co- somewhat higher than we found (30-day, 3.1% versus morbid conditions (all outcomes). Finally, we performed a 2.0%; 1-year, 17% versus 9.8%). These differences may sensitivity analysis stratified by the type of procedure (total represent a number of improvements, including greater 4Ciia ora fteAeia oit fNephrology of Society American the of Journal Clinical 94

Table 2. Events before and after parathyroidectomy

Before Parathyroidectomy (n=4435) After Parathyroidectomy (n=4435) Rate Ratio

Patients Rate per 100 Patients Median Rate per 100 After versus Outcome Events, Percent Events, Percent with Patient-Years with Follow-Up, Patient-Years Before n (95% CI) n (95% CI) Event, n (95% CI) Event, n yr (95% CI) (95% CI)

30-Day outcomes Hospitalization 1057 1251 0.08 23.8 344.4 (22.6 to 25.1) (325.2 to 363.6) ICU stay 367 367 0.08 8.3 105.8 (7.5 to 9.1) (93.6 to 115.2) Hospital days 1057 6259 0.08 23.8 1724.4 (22.6 to 25.1) (1681 to 1768) 1-Year outcomes Hospitalization 2477 6147 55.9 138.6 2832 7571 1.0 63.9 192.0 1.39 (54.4 to57.3) (135.1 to 142.1) (62.4 to 65.3) (187.7 to 196.3) (1.34 to 1.44) ICU stay 997 1680 22.5 37.9 1415 2524 1.0 31.9 64.0 1.69 required (21.3 to23.7) (36.1 to 39.7) (30.5 to 33.3) (61.5 to 66.5) (1.59 to 1.80) Total hospital 2477 35,004 789.3 2832 49,096 1.0 1245.0 1.58 days (781.0 to 797.5) (1234 to 1256) (1.56 to 1.60) Cause-specific hospitalization CHF 391 568 8.8 12.8 307 453 1.0 6.9 11.5 0.90 (8.0 to 9.7) (11.8 to 13.9) (6.2 to 7.7) (10.4 to 12.5) (0.80 to 1.02) CVA/TIA 67 76 1.5 1.7 101 124 1.0 2.3 3.1 (2.6 to 3.7) 1.83 (1.2 to 1.9) (1.3 to 2.1) (1.8 to 2.7) (1.38 to 2.43) Dysrhythmia 114 150 2.6 3.4 154 192 1.0 3.5 4.9 (4.2 to 5.6) 1.44 (2.1 to 3.0) (2.8 to 3.9) (2.9 to 4.0) (1.16 to 1.78) AMI 102 130 2.3 2.9 174 229 1.0 3.9 5.8 (5.1 to 6.6) 1.98 (1.9 to 2.7) (2.4 to 3.4) (3.4 to 4.5) (1.60 to 2.46) With 83 98 1.9 2.2 942 1486 1.0 21.2 37.7 17.05 hypocalcemia (1.5 to 2.3) (1.8 to 2.6) (20.0 to 22.4) (35.8 to 39.6) (13.90 to 20.92) Total outpatient visits Total ED/ 3309 13,941 74.6 314.3 3408 14,916 1.0 76.8 378.3 1.20 observation (73.3 to75.9) (309.1 to 319.6) (75.6 to 78.1) (372.2 to 384.3) (1.17 to 1.23) ED/observation 84 106 1.9 2.4 1072 1926 1.0 24.2 48.8 20.44 with hypocalcemia (1.5 to 2.3) (1.9 to 2.8) (22.8 to 25.4) (46.7 to 51.0) (16.81 to 24.85) Outpatient/physician 30 31 0.7 0.7 234 469 1.0 5.3 11.9 17.02 with hypocalcemia (0.4 to 0.9) (0.5 to 0.9) (4.6 to 5.9) (10.8 to 13.0) (11.83 to 24.48)

ICU, intensive care unit; CHF, congestive heart failur; AMI, acute myocardial infarction; ED, emergency department; 95% CI, 95% confidence interval. Clin J Am Soc Nephrol 10: 90–97, January, 2015 Clinical Outcomes after Parathyroidectomy, Ishani et al. 95

Figure 2. | Event rates in the 1 year before and 1 year after parathyroidectomy. (A) Total hospitalizations. (B) Total hospitalizations with hypocalcemia. (C) Total emergency department or observation visits. (D) Total emergency department or observation visits with hypocalcemia. RR and 95% CI values are based on the ratio of postparathyroidectomy to prior parathyroidectomy. 95% CI, 95% confidence interval; ED, emergency department; PTx, parathyroidectomy; RR, rate ratio. surgical experience, improved surgical technique, or im- number of cost-effectiveness studies have attempted to proved patient selection. Alternatively, they may arise compare surgical parathyroidectomy with other therapeutic from how the two cohorts were constructed; specifically, interventions (20–22). These studies have generally we excluded patients with ,1 year of dialysis before para- overlooked a significant source of costs for the parathy- thyroidectomy. This exclusion may have biased our re- roidectomy group, specifically increased outpatient utili- sults toward a healthier population. Given our present zation and higher risk of hospitalization within the first 30 data, we are unable to draw definitive conclusion regard- days and the first year. An analysis by Belozeroff et al. (23) ing temporal changes in survival after parathyroidec- suggested that after accounting for a decrease in drug tomy. Finally, Rudser et al. (18) evaluated the risk of utilization and an increase in physician encounters, total fractures after parathyroidectomy in a nationwide cohort, costs after parathyroidectomy increased by $434 per and demonstrated a lower risk of fractures (hip, vertebra, month. Future analyses assessing costs of management and distal radius-wrist) in patients who underwent para- of elevated PTH should account for the increased medical thyroidectomy compared with matched controls. utilization after parathyroidectomy and patient quality of In the current analysis, we used a pre-post comparison life, in addition to the benefits that have been attributed to to investigate the risks related to parathyroidectomy. parathyroidectomy. This was primarily motivated by the knowledge that Our study should be evaluated in light of the following patients who ultimately undergo parathyroidectomy are limitations. First, our results included only Medicare highly selected (e.g., assessed to be able to survive the participants undergoing in-center hemodialysis for at least 1 ), and secondly by the understanding that the di- year before parathyroidectomy. The generalizability of our rect patient outcomes should be viewed from the per- results to other populations is unknown, although it is spective of how the patients will contrast their care possible that broader application of parathyroidectomy before and after the surgery. Previous studies have at- would lead to greater use in less appropriate patients, tempted to contrast patients who do and do not undergo potentially worsening the risk/benefitratio.Second,the parathyroidectomy and control for potential differences effect of parathyroidectomy on risk of mortality is an im- through individual or propensity score matching; with portant clinical question, but we were unable to evaluate it such an approach, substantial residual bias likely re- given our study design. Next, we were unable to determine mains because all prognostic factors that physicians whether the adverse outcomes observed in the postpara- may assess when deciding whether to perform parathy- thyroidectomy period directly resulted from the parathy- roidectomy are unlikely to be recorded in databases, and roidectomy procedure, whether patients with ESRD in thus the validity of these previous results seems question- general fare poorly after surgery (24), or whether the out- able. Therefore, it is not surprising that these studies have comes were simply due to receiving maintenance dialysis found that survival in patients selected for parathyroidec- for an additional year (25). Given the consistency of results tomy is better than in a matched population. We were un- across time in dialysis subgroups, the latter seems an un- able to directly assess the potential survival benefitof likely explanation. Finally, because we followed patients parathyroidectomy with our study design. We did observe for only 1 year, we were unable to evaluate the long-term a 30-day mortality of 2%, similar to other dialysis studies outcomes after parathyroidectomy. (5,19), but substantially higher than in studies evaluating Our study also has a number of strengths. We used a 30-day mortality after parathyroidectomy in a mixed general nationwide cohort of patients receiving dialysis who population (0.11% at 30 days) (2). underwent parathyroidectomy during a contemporary Given the presumed benefits of parathyroidectomy period. Because of the pre-post design, our results are on both laboratory parameters and clinical outcomes, a not limited by selection bias. Finally, our study evaluated 96 Clinical Journal of the American Society of Nephrology

Figure 3. | Outcomes by select patient characteristics. Data points in gray represent subgroups in which a significant interaction (P,0.05) exists between the subgroup and parathyroidectomy. Data points in black have a nonsignificant P for interaction (P.0.05). ASHD, atherosclerotic heart disease; BMI, body mass index; CHF, congestive heart failure; CVA/TIA, cerebrovascular accident/transient ischemic attack; ICU, intensive care unit; PVD, peripheral vascular disease. patient-oriented outcomes and can serve as the basis for Acknowledgments counseling patients regarding outcomes should they elect to The authors thank Chronic Disease Research Group colleagues pursue parathyroidectomy. Delaney Berrini for manuscript preparation, Edward Constantini Parathyroidectomy is associated with significant morbidity, and Susan Everson for figure design, and Nan Booth, MSW, ELS, for including more hospitalizations, ICU stays, total hospital days, manuscript editing. and emergency department/observation visits, both with and This study was supported by a research contract from Amgen Inc without hypocalcemia treatment. No change was detected in (Thousand Oaks, CA). The contract provides for the Minneapolis fracture rates up to 1 year after parathyroidectomy, given that Medical Research Foundation authors to have final determination of fractures occurred infrequently. Mortality during the para- manuscript content. thyroidectomy hospitalization and the 30 days immedi- The data reported here were supplied by the USRDS. The in- ately after discharge was substantial at 2%. Although we terpretation and reporting of these data are the responsibility of the are unable to provide a comprehensive risk-benefit analysis authors and in no way should be seen as an official policy or in- of parathyroidectomy with the present data, the new in- terpretation of the US Government. formation we report contributes to the understanding of the risks involved, assisting providers and patients in making Disclosures informed decisions. Future work such as investigating select A.I., J.L., J.B.W., and A.J.C. are employed by the Chronic Disease patient groups with particularly severe disease or suboptimal Research Group, which receives research support from Amgen. outcomes should be undertaken to better understand the K.A.L., T.D., and B.D.B. are employed by Amgen. G.A.B. is employed risks and benefits of parathyroidectomy in dialysis patients by Denver Nephrologists, which receives research support from with SHPT. Amgen. Clin J Am Soc Nephrol 10: 90–97, January, 2015 Clinical Outcomes after Parathyroidectomy, Ishani et al. 97

References Chen SC, Daniels F, Ebben J, Frazier E, Johnson R, Sheets D, Wang 1. Tentori F, Blayney MJ, Albert JM, Gillespie BW, Kerr PG, Bommer X, Forrest B, Berrini D, Constantini E, Everson S, Eggers P, Agodoa J, Young EW, Akizawa T, Akiba T, Pisoni RL, Robinson BM, Port L: US Renal Data System 2013 Annual Data Report. Am J Kidney FK: Mortality risk for dialysis patients with different levels of serum Dis 63[Suppl]: A7, 2014 calcium, phosphorus, and PTH: The Dialysis Outcomes and Prac- 15. Kestenbaum B, Seliger SL, Gillen DL, Wasse H, Young B, Sherrard tice Patterns Study (DOPPS). Am J Kidney Dis 52: 519–530, 2008 DJ, Weiss NS, Stehman-Breen CO: Parathyroidectomy rates 2. Gupta PK, Smith RB, Gupta H, Forse RA, Fang X, Lydiatt WM: among United States dialysis patients: 1990-1999. Kidney Int 65: Outcomes after thyroidectomy and parathyroidectomy. Head 282–288, 2004 Neck 34: 477–484, 2012 16. Li S, Chen YW, Peng Y, Foley RN, St Peter WL: Trends in para- 3. Kidney Disease: Improving Global Outcomes (KDIGO) CKD- thyroidectomy rates in US hemodialysis patients from 1992 to MBD Work Group: KDIGO clinical practice guideline for the 2007. Am J Kidney Dis 57: 602–611, 2011 diagnosis, evaluation, prevention, and treatment of Chronic 17. Chuang CH, Wang JJ, Weng SF, Chung KM, Chen YP, Huang CC, Kidney Disease-Mineral and Bone Disorder (CKD-MBD). Kidney Yang CM, Chien CC: Epidemiology and mortality among dialysis Int Suppl 113: S1–S130, 2009 patients with parathyroidectomy: Taiwan National Cohort Study. 4. National Kidney Foundation: K/DOQI clinical practice guide- J Nephrol 26: 1143–1150, 2013 lines for bone metabolism and disease in . 18. Rudser KD, de Boer IH, Dooley A, Young B, Kestenbaum B: Am J Kidney Dis 42[Suppl 3]: S1–S201, 2003 Fracture risk after parathyroidectomy among chronic hemodial- 5. Kestenbaum B, Andress DL, Schwartz SM, Gillen DL, Seliger SL, ysis patients. J Am Soc Nephrol 18: 2401–2407, 2007 Jadav PR, Sherrard DJ, Stehman-Breen C: Survival following 19. Sharma J, Raggi P, Kutner N, Bailey J, Zhang R, Huang Y, Herzog parathyroidectomy among United States dialysis patients. Kidney CA, Weber C: Improved long-term survival of dialysis patients Int 66: 2010–2016, 2004 after near-total parathyroidectomy. J Am Coll Surg 214: 400–407, 6. Trombetti A, Stoermann C, Robert JH, Herrmann FR, Pennisi P, discussion 407–408, 2012 Martin PY, Rizzoli R: Survival after parathyroidectomy in patients 20. Narayan R, Perkins RM, Berbano EP, Yuan CM, Neff RT, Sawyers with end-stage renal disease and severe hyperparathyroidism. ES, Yeo FE, Vidal-Trecan GM, Abbott KC: Parathyroidectomy World J Surg 31: 1014–1021, 2007 versus cinacalcet hydrochloride-based medical therapy in the 7. Naranda J, Ekart R, Pecovnik-Balon B: Total parathyroidectomy management of hyperparathyroidism in ESRD: a cost utility with forearm autotransplantation as the treatment of choice for analysis. Am J Kidney Dis 49: 801–813, 2007 secondary hyperparathyroidism. JIntMedRes39: 978–987, 21. Plosker GL: Cinacalcet: A pharmacoeconomic review of its use 2011 in secondary hyperparathyroidism in end-stage renal disease. 8. Kovacevic B, Ignjatovic M, Zivaljevic V, Cuk V, Scepanovic M, Pharmacoeconomics 29: 807–821, 2011 Petrovic Z, Paunovic I: Parathyroidectomy for the attainment of 22. Komaba H, Moriwaki K, Goto S, Yamada S, Taniguchi M, Kakuta NKF-K/DOQI™ and KDIGO recommended values for bone and T, Kamae I, Fukagawa M: Cost-effectiveness of cinacalcet hy- mineral metabolism in dialysis patients with uncontrollable secondary drochloride for hemodialysis patients with severe secondary hyperparathyroidism. Langenbecks Arch Surg 397: 413–420, 2012 hyperparathyroidism in Japan. Am J Kidney Dis 60: 262–271, 9. Gagne´ ER, Urena~ P,Leite-Silva S, Zingraff J, Chevalier A, Sarfati E, 2012 Dubost C, Dru¨eke TB: Short- and long-term efficacy of total 23. Belozeroff V, Cooper K, Hess G, Chang CL: Healthcare use and parathyroidectomy with immediate autografting compared with costs before and after parathyroidectomy in patients on dialysis. subtotal parathyroidectomy in hemodialysis patients. JAmSoc BMC Health Serv Res 13: 248, 2013 Nephrol 3: 1008–1017, 1992 24. Schneider CR, Cobb W, Patel S, Cull D, Anna C, Roettger R: 10. Gasparri G, Camandona M, Abbona GC, Papotti M, Jeantet A, Elective surgery in patients with end stage renal disease: what’s Radice E, Mullineris B, Dei Poli M: Secondary and tertiary hy- the risk? Am Surg 75: 790–793, discussion 793, 2009 perparathyroidism: Causes of recurrent disease after 446 para- 25. Chertow GM, Johansen KL, Lew N, Lazarus JM, Lowrie EG: . Ann Surg 233: 65–69, 2001 Vintage, nutritional status, and survival in hemodialysis patients. 11. Neonakis E, Wheeler MH, Krishnan H, Coles GA, Davies F, Kidney Int 57: 1176–1181, 2000 Woodhead JS: Results of surgical treatment of renal hyperpara- thyroidism. Arch Surg 130: 643–648, 1995 Received: April 7, 2014 Accepted: September 29, 2014 12. Jofre´ R, Lo´pez Go´mez JM, Mena´rguez J, Polo JR, Guinsburg M, Villaverde T, Pe´rez Flores I, Carretero D, Rodrı´guez Benitez P, Published online ahead of print. Publication date available at www. Pe´rez Garcı´a R: Parathyroidectomy: Whom and when? Kidney Int cjasn.org. Suppl 85: S97–S100, 2003 13. Saunders BD, Wainess RM, Dimick JB, Doherty GM, Upchurch This article contains supplemental material online at http://cjasn. GR, Gauger PG: Who performs endocrine operations in the United States? Surgery 134: 924–931, discussion 931, 2003 asnjournals.org/lookup/suppl/doi:10.2215/CJN.03520414/-/ 14. Collins AJ, Foley RN, Chavers B, Gilbertson D, Herzog C, Ishani A, DCSupplemental. Johansen K, Kasiske BL, Kutner N, Liu J, St Peter W, Guo H, Hu Y, Kats A, Li S, Li S, Maloney J, Roberts T,Skeans M, Snyder J, Solid C, See related editorial, “Dysphoria Induced in Dialysis Providers by Thompson B, Weinhandl E, Xiong H, Yusuf A, Zaun D, Arko C, Secondary Hyperparathyroidism,” on pages 9–11. Supplemental Appendix S1. Fracture Definitions from Claims

Fractures studied were those of the pelvis/hip, femur, lower leg, forearm, shoulder/upper

arm, and rib/sternum. They were defined from Medicare Part B claims. Both International

Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes

and Current Procedural Terminology (CPT) procedure codes were required for defining

pelvis/hip, femur, lower leg, forearm, and shoulder/upper arm fractures (i.e., both the diagnosis

code and the procedure code had to appear on the same line in the claim). Additionally, for

pelvis/hip and femur fractures, the place of service was required to be either the inpatient setting

or the emergency department. Fractures of the rib/sternum did not require a CPT code for

diagnosis because they rarely require surgical intervention. Multiple fractures were counted, but

a fracture claim was considered a “new” fracture only when it occurred at least 90 days after a

previous claim for the same fracture (to minimize misclassification of complications of previous

fracture events as new events). The ICD-9-CM diagnosis and CPT procedure codes used are

listed in the Supplemental Table S1.

Supplemental Table S1. Codes used to identify fractures from billing claims

Additional requirements Fracture Site ICD-9-CM Diagnosis Codes CPT Codes for Part B Claims Pelvis/hip 733.14, 808.xx, 820.xx 27193-27248 Place of service IP or ED Femur 733.15, 821.xx 27500-27514, 27520-27540, Place of service IP or ED 29850-29856 Lower leg 733.16, 822.xx, 823.xx, 824.xx 27750-27828 – Forearm 813.xx, 814.xx 25500-25526, 25530-25609, – 25622-25652, 25680-25695 Shoulder/upper arm 733.11, 810.xx-812.xx 23500-23515, 23570-23630, – 23665-23680, 24500-24587, 24620-24685 (except 24640) Rib/sternum 807.0-807.3 – –

CPT, Current Procedural Terminology; ED, emergency department; ICD-9-CM, International

Classification of Diseases, Ninth Revision, Clinical Modification; IP, inpatient. Supplemental Figure S1.

Hospitaliza on ICU Admission ED/Observa on Stays

All

19-44

e 45-64 g

A 65-74 (years) 75+

e White c

a Black R Other

Male

Sex Female

y

e r Diabetes D s a u R Hypertension m a S i E r C Glomerulonephri s P Other/Unknown/Missing

<18 ) 2 18-<25 25-<30 30-<35 I (kg/m 35-<40 M

B ≥40 Missing e ) g

r 1-<3 a a t

e 3-<5 Y ( ≥5 Vi n

CHF CHF No Yes

CVA/TIA No Yes s n o ASHD No i Yes d n o c No d PVD i

b Yes r m o

o Dysrhythmia No C Yes

Cardiac disease, No other Yes

Diabetes No Yes

-20 0 20 40 60 80 100 120 140 -10 0 10 20 30 40 50 60 -40 -20 0 20 40 60 80 100 120 140 Rate Difference (95% CI) Rate Difference (95% CI) Rate Difference (95% CI)