Mary Kay Theis

Acute Mountain Sickness in a General Tourist Population at Moderate Altitudes

Rapid ascent from low to high altitude is often followed by headache, fatigue, shortness of breath, sleeplessness, and anorexia, a symptom complex called acute mountain sickness. Although some of these symptoms may occur as a result of travel not associated with altitude, only 5% of adults traveling at sea level report similar symptoms [1]. A long-standing interest has existed in the study of acute mountain sickness because it affects a large number of mountain visitors [2-4] and can progress to the life-threatening conditions of high-altitude pulmonary edema or high-altitude cerebral edema [5]. Previous estimates of the incidence of acute mountain sickness have been obtained primarily from small groups of physically fit young men going to altitudes above 12 000 feet [2-4, 7-9]. Little information exists on the frequency and severity of the disorder in the general population at moderate altitudes, yet the population at risk is large. For example, more than 13 million persons visited the Colorado mountains in 1990 for business, conferences, or recreational activities including skiing, climbing, hiking, hunting, and fishing [10]. More needs to be learned about the incidence of acute mountain sickness at moderate altitudes in the general population and about the characteristics of those most likely to be at risk for symptom development. We therefore surveyed groups of persons visiting resorts in the Colorado mountains for conferences and seminars. Specifically, we sought to determine 1) the incidence of acute mountain sickness in visitors exposed to moderate elevations; 2) the effect of acute mountain sickness on physical activity; and 3) the visitor characteristics associated with the development of acute mountain sickness. This information would be useful for developing strategies to minimize symptoms in travelers to moderate altitudes. Methods The study cohort consisted of 4212 adults attending 45 conferences at resorts located at elevations of 6300 to 9700 feet in the Rocky Mountains of Colorado from July 1989 to May 1991. Resorts were chosen on the basis of the willingness of conference organizers to participate. Conferences whose schedules required all participants to attend a meeting within 48 hours of arrival when the study questionnaire could be distributed were included. Study personnel attended these meetings, briefly introduced the study, and distributed the questionnaires. Questions by participants concerning acute mountain sickness or the effects of altitude on health were not answered until all questionnaires were collected. Completion of the survey usually took less than 10 minutes. The participants in each meeting were counted to calculate the response rate. The questionnaire was completed by 3158 (75%) of the persons registered for these conferences, and information satisfactory for analysis was obtained from 99% of those completed. Visitors ranged in age from 16 to 87 years (mean age [SD], 657e43.8 11.8 years) (Figure 1). Of the completed surveys, 2023 (64%) were conducted at resorts located at elevations over 9000 feet and 2603 (82%) were completed in the winter season (November through April). The study was approved by the Human Research Committee of the University of Colorado Health Sciences Center. Figure 1. Distribution of visitors by age and physical condition. n n The questionnaire was designed to obtain demographic information concerning age, gender, height, weight, and permanent residence for each visitor; previous or current medications; and the duration of stops, if any, made en route. Questions asked regarding underlying health conditions included whether the participant had ever been treated for lung disease (asthma, bronchitis, or emphysema); heart conditions (angina or heart attacks); diabetes; high blood pressure; or pregnancy. Participants were asked How do you rate your physical condition? Responses were rated as great, good, average, or poor. To determine whether participants had acute mountain sickness, they were asked if they had experienced any of the following symptoms while at the resort: loss of appetite, vomiting, shortness of breath, dizziness or lightheadedness, unusual fatigue, sleep disturbance (other than related to normal travel), and headache. If the response to headache was yes, they were asked to describe it as mild or severe. Acute mountain sickness was defined as the presence of three or more of these symptoms in the setting of recent altitude exposure. This case definition was similar to that used by previous investigators [1, 2, 5, 6, 9, 12, 13] and is in accordance with the case definition recently developed and codified for international use by the International Hypoxia Symposium [14]. If participants had any of these symptoms of acute mountain sickness, they were asked to determine how the symptoms affected their activity. Response options included no limitation, reduced activity, and required to stay in bed or room. A determination of symptom onset was assessed by asking participants How long after arrival at the resort did symptoms begin? Response options included less than 12 hours, 12 to 24 hours, 25 to 48 hours, 49 to 72 hours, or 72 hours. Alcohol use was measured as the number of beer, wine, or hard-liquor drinks consumed within the first 24 hours of arrival at the resort. Season was determined by the date of questionnaire administration, with winter defined as November through April in each of the study years. Body mass index was calculated as weight kg/in2 and was used to identify obese persons (women with a body mass index >27.3 and men with a body mass index >27.8) [11]. Persons with or without acute mountain sickness were compared using the Wilcoxon rank-sum test for ordinal variables and using the chi-square test for categoric variables. The Student t-test was used for normally distributed variables. The Fisher exact test was used for small sample sizes. Associations were considered significant at P < 0.05. A forward, stepwise, multiple logistic regression analysis was used to examine the independent effects of participant characteristics on the occurrence of acute mountain sickness. All variables associated with the occurrence of acute mountain sickness at P < 0.25 were initially included in the regression analysis. Data acquired in year 2 (June 1990 to May 1991), comprising 1241 cases after the revision of the questions concerning underlying health conditions and habitual activity level before travel, were used for the regression analysis. Variables were dichotomized for ease in presentation. The adjusted odds ratios were computed with 95% confidence intervals (CIs). All calculations were done using the Statistical Analysis Systems statistical package (Cary, North Carolina) [15]. Results Most of the visitors were middle-aged men whose permanent addresses were at sea level, who did not smoke, and who considered themselves to be in good physical condition (Table 1; Figure 1). Approximately one third (28%) stopped overnight at an intermediate altitude (5280 feet) enroute to their destination. Most (64%) had consumed one or more alcoholic beverages in the first 24 hours after arrival. Small proportions of the visitors were obese, pregnant, or had chronic illnesses (Table 1). Table 1. Characteristics and Incidence of Acute Mountain Sickness in Visitors to Areas of Moderate Altitude* Twenty-five percent (CI, 24.98% to 25.01%) of the visitors reported having three or more symptoms and thus met the case definition for acute mountain sickness, whereas 73% had at least one reported symptom. The most common symptom was headache, and the least common was vomiting (Figure 2). For most participants (65%), the onset of symptoms occurred within the first 12 hours after arrival at altitude; symptom onset occurred between 12 and 36 hours in 34% and after more than 36 hours in 1%. Most (58%) of those with symptoms took analgesics (for example, aspirin, acetaminophen, or ibuprofen). Although 44% of persons with acute mountain sickness had no reduction in activity, 51% had moderate activity reduction, and a small proportion (5%) stayed in bed. Figure 2. Distribution of symptoms of acute mountain sickness in 3072 visitors. Visitors whose permanent residence was at an elevation below 3000 feet were more likely to develop acute mountain sickness (see Table 1). The frequency with which it developed was inversely related to age and physical condition (Figure 3). Altitude visited and a previous history of acute mountain sickness were associated with an increased occurrence, whereas development was inversely related to alcohol consumption. Visitors who stopped over at lower elevations for more than 38 hours were less likely to develop acute mountain sickness than were those who did not. Obesity, female gender, and chronic lung disease were also associated with the development of acute mountain sickness. Figure 3. Percentage of acute mountain sickness in visitors to moderate altitudes according to age, physical condition, and altitude visited. n P n P n P The following nine variables were entered into the regression analysis as dichotomous variables: age (younger than 60 years), sex, altitude of permanent residence (below 3000 feet), obesity, lung disease, diabetes, overnight stops before arrival at the resort, previous symptoms during past altitude travel, and self-reported physical condition (poor or average). Although alcohol was statistically associated with acute mountain sickness, it was not included in the model because of the inability to exclude a temporal effect (that is, participants may have become sick and subsequently decided not to drink). The five independent predictors of acute mountain sickness, based on the logistic regression, were residence at an altitude less than 3000 feet; symptoms of acute mountain sickness during previous altitude travel; age younger than 60 years; physical condition self-assessed as poor or average; and the presence of lun

Comparative effectiveness of bariatric surgery vs. nonsurgical treatment of type 2 diabetes among severely obese adults

Although all weight-loss approaches may improve insulin sensitivity in type 2 diabetes, bariatric surgery is believed to be the only reliable means of achieving diabetes remission. We conducted a retrospective cohort study to compare rates of diabetes remission, relapse and all-cause mortality among severely obese individuals with diabetes who underwent bariatric surgery vs. nonsurgically treated individuals. Severely obese adults with uncontrolled or medication-controlled diabetes who underwent bariatric surgery or received usual medical care from 2005 to 2008 in three health care delivery systems in the United States were eligible. Diabetes status was identified using pharmacy, laboratory, and diagnosis information from electronic medical records. A propensity approach and exclusion criteria identified 1395 adults with diabetes who had bariatric surgery and 62,322 who did not. Most procedures were Roux-en-Y gastric bypass (72.0% laparoscopic; 8.2% open); 4.4% were gastric banding, 2.4% sleeve gastrectomy, and 13.2% were other procedures. At two years, bariatric subjects experienced significantly higher diabetes remission rates [73.7% (95% CI: 70.6, 76.5)] compared to nonsurgical subjects [6.9% (95%CI: 6.9, 7.1)]. Age, site, duration of diabetes, hemoglobin A1c level, and intensity of diabetes medication treatment were significantly associated with remission. Bariatric subjects also experienced lower relapse rates than nonsurgical subjects (adjusted HR: 0.19; 95% CI: 0.15-0.23) with no higher risk of death (adjusted HR: 0.54; 95% CI: 0.22-1.30). We conclude that bariatric surgery can effectively induce remission of diabetes among most severely obese adults, and this treatment approach appears to be superior to nonsurgical treatment in inducing diabetes remission.

Long-term Microvascular Disease Outcomes in Patients With Type 2 Diabetes After Bariatric Surgery: Evidence for the Legacy Effect of Surgery.

OBJECTIVE To identify and quantify any legacy effect of bariatric surgery on risk of incident microvascular disease in patients with type 2 diabetes. RESEARCH DESIGN AND METHODS We conducted a retrospective observational cohort study (n = 4,683; 40% racial/ethnic minority) of patients with type 2 diabetes who underwent bariatric surgery from 2001 through 2011. The primary outcome measure was incident microvascular disease defined as a composite indicator of the first occurrence of retinopathy, neuropathy, and/or nephropathy. The Cox proportional hazards framework was used to investigate the associations between type 2 diabetes remission/relapse status and time to microvascular disease. RESULTS Covariate-adjusted analyses showed that patients who experienced type 2 diabetes remission had 29% lower risk of incident microvascular disease compared with patients who never remitted (hazard ratio [HR] 0.71 [95% CI 0.60, 0.85]). Among patients who experienced a relapse after remission, the length of time spent in remission was inversely related to the risk of incident microvascular disease; for every additional year of time spent in remission prior to relapse, the risk of microvascular disease was reduced by 19% (HR 0.81 [95% CI 0.67, 0.99]) compared with patients who never remitted. CONCLUSIONS Our results indicate that remission of type 2 diabetes after bariatric surgery confers benefits for risk of incident microvascular disease even if patients eventually experience a relapse of their type 2 diabetes. This provides support for a legacy effect of bariatric surgery, where even a transient period of surgically induced type 2 diabetes remission is associated with lower long-term microvascular disease risk.

A Population-Based Study of Periodontal Care Among Those With and Without Diabetes

BACKGROUND Our objectives were to describe the prevalence of periodontal care (a marker of periodontitis) among persons with and without diabetes and to examine the association between periodontal care and diabetes. METHODS We conducted a cross-sectional analysis, using 5 years of electronic data from a population-based cohort (N = 46,132), aged 40 to 70 years, with dental and medical insurance, and ≥ 1 dental and ≥ 1 medical visit. Periodontal care (yes/no) was defined by dental claims codes for procedures used to manage periodontitis. The association between periodontal care and diabetes was determined using logistic regression adjusted for and stratified by age, sex, insurance type, smoking status, body mass index (BMI) (in kilograms per square meter), and resource utilization band (RUB) (a measure of expected health care utilization attributable to comorbidity). RESULTS Overall, 11.2% (5,153 of 46,132) met diabetes criteria. The age-adjusted prevalence of periodontal care among those with and without diabetes was 39.1% and 32.5%, respectively (P <0.0001). The association between diabetes and periodontal care decreased with increasing age (interaction, P <0.0001), adjusting for BMI and RUB. The aged-stratified, adjusted odds ratio (OR) for periodontal care associated with diabetes was highest among those aged 40 to 44 years [OR, 1.6; confidence interval (CI), 1.30 to 1.97] and lowest among those aged 60 to 64 years (OR, 0.97; CI, 0.81 to 1.15) and was significant only among those aged 40 to 54 years. CONCLUSION We found that the prevalence of periodontal care was significantly higher among those with diabetes compared to those without diabetes and that the magnitude of this association decreased with increasing age.

Influence of body mass index on the choice of therapy for depression and follow-up care

Objective: Overweight and obese patients commonly suffer from depression and choice of depression therapy may alter weight. We conducted a cohort study to investigate whether obesity is associated with treatment choices for depression; and whether obesity is associated with appropriate duration of depression treatment and receipt of follow‐up visits.

Antipsychotic medications and extreme weight gain in two health systems

OBJECTIVE We conducted this study to investigate the rate of clinically important, extreme weight gain (EWG; ≥7% body weight gain) among all second generation antipsychotic (SGA) users in two large health care systems in the United States. STUDY DESIGN Retrospective observational cohort study. METHODS We used electronic medical record databases of two health systems to identify adults aged 18-79 years who from 1 January 2004 to 31 December 2011 had initiated a SGA medication. All patients had to have a minimum of two weight measures in the medical record: (1) one or more weights in the 180-day pre-treatment (baseline) period; and (2) one or more weights in the first year after initiating SGA treatment. RESULTS We found that EWG occurred in 7.7-17.0% of SGA users. At one year, the average weight gain was nearly 10kg among SGA users who experienced EWG. Olanzapine was the SGA most commonly associated with EWG with a rate of 17.0 per 100 users [95% confidence interval (CI): 14.2-20.5], while ziprasidone was least commonly associated with EWG (7.7 per 100 users; 95% CI: 4.6-13.0). CONCLUSIONS We found that clinically-important weight gain was common after the initiation of SGA treatment, and the EWG phenotype was easily identifiable within electronic medical records. There was significant heterogeneity in the rate of EWG across SGA medications. Weight gains of this magnitude are likely to have adverse health consequences and there is a significant unmet opportunity for physicians to identify these events and mitigate the harms of SGA use.

Long-Term Weight Change after Initiating Second-Generation Antidepressants

(1) Objective: To examine the relationship between the choice of second-generation antidepressant drug treatment and long-term weight change; (2) Methods: We conducted a retrospective cohort study to investigate the relationship between choice of antidepressant medication and weight change at two years among adult patients with a new antidepressant treatment episode between January, 2006 and October, 2009 in a large health system in Washington State. Medication use, encounters, diagnoses, height, and weight were collected from electronic databases. We modeled change in weight and BMI at two years after initiation of treatment using inverse probability weighted linear regression models that adjusted for potential confounders. Fluoxetine was the reference treatment; (3) Results: In intent-to-treat analyses, non-smokers who initiated bupropion treatment on average lost 7.1 lbs compared to fluoxetine users who were non-smokers (95% CI: −11.3, −2.8; p-value < 0.01); smokers who initiated bupropion treatment gained on average 2.2 lbs compared to fluoxetine users who were smokers (95% CI: −2.3, 6.8; p-value = 0.33). Changes in weight associated with all other antidepressant medications were not significantly different than fluoxetine, except for sertraline users, who gained an average of 5.9 lbs compared to fluoxetine users (95% CI: 0.8, 10.9; p-value = 0.02); (4) Conclusion: Antidepressant drug therapy is significantly associated with long-term weight change at two years. Bupropion may be considered as the first-line drug of choice for overweight and obese patients unless there are other existing contraindications.

Learning About Missing Data Mechanisms in Electronic Health Records-based Research: A Survey-based Approach.

Background: Bias due to missing data is a major concern in electronic health record (EHR)-based research. As part of an ongoing EHR-based study of weight change among patients treated for depression, we conducted a survey to investigate determinants of missingness in the available weight information and to evaluate the missing-at-random assumption. Methods: We identified 8,345 individuals enrolled in a large EHR-based health care system who had monotherapy treatment for depression from April 2008 to March 2010. A stratified sample of 1,153 individuals completed a detailed survey. Logistic regression was used to investigate determinants of whether a patient (1) had an opportunity to be weighed at treatment initiation (baseline), and (2) had a weight measurement recorded. Parallel analyses were conducted to investigate missingness during follow-up. Throughout, inverse-probability weighting was used to adjust for the design and survey nonresponse. Analyses were also conducted to investigate potential recall bias. Results: Missingness at baseline and during follow-up was associated with numerous factors not routinely collected in the EHR including whether or not the patient had ever chosen not to be weighed, external weight control activities, and self-reported baseline weight. Patient attitudes about their weight and perceptions regarding the potential impact of their depression treatment on weight were not related to missingness. Conclusion: Adopting a comprehensive strategy to investigate missingness early in the research process gives researchers information necessary to evaluate key assumptions. While the survey presented focuses on outcome data, the overarching strategy can be applied to any and all data elements subject to missingness.

Microvascular Outcomes in Patients With Diabetes After Bariatric Surgery Versus Usual Care: A Matched Cohort Study

The profound effects of bariatric surgery on glycemic control are well established in many studies, including several randomized trials that compared surgery versus intensive medical and lifestyle treatment of type 2 diabetes mellitus (T2DM) (17). Remission of T2DM is common after bariatric surgery, and Roux-en-Y gastric bypass (RYGB) outperforms sleeve gastrectomy (SG) in most reports (4, 8, 9). More important than improvements in glycemic control is whether these glycemic changes reduce the incidence of major complications, such as microvascular events. In the United States, T2DM is the primary cause of nephropathy, end-stage renal disease, adult blindness, and neuropathy, which together lead to significant downstream costs and decreased quality of life (1012). Thus, a major goal of diabetes treatment is to mitigate the risk for these long-term sequelae (13). We have previously shown that in patients with T2DM who have bariatric surgery, risk for incident microvascular disease is reduced for every year of remission from T2DMeven if patients eventually relapse back to T2DM (indicating a legacy effect) (14). However, a detailed 2016 review of long-term microvascular outcomes of diabetes concluded that data were inadequate to support a superior effect of bariatric surgery compared with medical therapy (15). Two recently published studies help inform this question (4, 16). The STAMPEDE (Surgical Therapy and Medications Potentially Eradicate Diabetes Efficiently) trial (4) randomly assigned 150 patients with obesity and T2DM to have RYGB, SG, or an intensive medical and lifestyle intervention. After 5 years, urinary albumincreatinine ratios had decreased in the SG group only. Retinopathy did not change in any group, and neuropathy was not examined. STAMPEDE was limited by a relatively small sample size, which made it underpowered for rare outcomes. The prospective SOS (Swedish Obesity Subjects) matched cohort study (16) followed 2010 patients who had bariatric surgery (13% RYGB, 19% gastric banding, and 68% vertical-banded gastroplasty) and 2037 matched patients who received usual care. The incidence of microvascular disease after 15 years of follow-up was significantly lower in the surgery group than the control group (hazard ratio [HR], 0.56 [95% CI, 0.48 to 0.66]), including among subgroups with prediabetes (HR, 0.18 [CI, 0.11 to 0.30]), established diabetes (HR, 0.54 [CI, 0.40 to 0.72]), and normoglycemia (HR, 0.63 [CI, 0.48 to 0.81]) at baseline. The SOS study was limited by the small number of RYGB procedures and lack of SG procedures, which are now the 2 most common bariatric procedures worldwide (17). To address these gaps in the literature, we did a matched cohort study to determine whether bariatric surgery was associated with lower incidence of microvascular disease than usual care for severe obesity and T2DM. We hypothesized that patients having bariatric surgery would have lower risk for any incident microvascular disease (composite of the first occurrence of retinopathy, neuropathy, or nephropathy). We further hypothesized that patients having bariatric surgery would have lower risk for incident retinopathy, nephropathy, and neuropathy than nonsurgical patients. Methods Settings We did a retrospective observational cohort study of adults with T2DM who had bariatric surgery between 2005 and 2011 while enrolled in 1 of 4 integrated health care systems from the Health Care Systems Research Network: Kaiser Permanente (KP) Washington in Washington state, HealthPartners in Minnesota, KP Northern California, and KP Southern California. All study procedures were reviewed and approved by the institutional review board at each site, and we were permitted to do the research without explicit consent from participants. Data Sources At each study site, staff used standardized electronic medical records, insurance claims, and other data systems (18) to extract enrollee data, including enrollment and insurance coverage; demographics; blood pressure; height; weight; laboratory values; medications dispensed; deaths; outpatient, inpatient, and emergency department use; and inpatient and outpatient diagnosis and procedure codes. Surgical Participants The bariatric population included adults (aged 19 to 79 years) with severe obesity (body mass index [BMI] 35 kg/m2) and T2DM who had a primary (first observed) bariatric surgical procedure between 1 January 2005 and 31 December 2011. Following an approach adopted in prior studies (14, 19, 20), we used a combination of bariatric registries; review of medical records; codes from the International Classification of Diseases, Ninth Revision (ICD-9) (43.89, 44.31, 44.38, 44.39, 44.68, 44.69, and 44.95); and Current Procedural Terminology codes (43633, 43644, 43645, 43659, 43770, 43775, 43842, 43843, 43844, 43845, 43846, and 43847) to identify bariatric procedures. Patients were classified as having T2DM if they had a hemoglobin A1c (HbA1c) level of at least 6.5% or fasting plasma glucose level of at least 6.99 mmol/L (126 mg/dL) at the most recent measurement within 2 years before surgery, or a current prescription for any diabetes medication. After selecting the population, we applied the following exclusion criteria based on information in the 2 years before surgery: less than 1 full year of continuous enrollment; history of gastrointestinal surgery for cancer; pregnancy in the year before surgery; gestational diabetes as the sole diabetes diagnosis; metformin as the sole indicator of possible T2DM (that is, no other T2DM medications, laboratory values, or diagnoses); preexisting neuropathy, retinopathy, or nephropathy (defined in Outcome and Censoring Definitions); and maximum preoperative BMI less than 35 kg/m2. This left 4972 patients. We further excluded 850 patients who were missing values for BMI, HbA1c level, or serum creatinine concentration in the 2 years before surgery. Surgical patients who were missing these values differed from those who had complete data (Appendix Table 1). Specifically, they were more likely to be non-Hispanic white, to be from the HealthPartners site, to have had their procedure in the first year of our study (2005), and to have had a shorter observed duration of T2DM at the time of surgery. Of the remaining 4122 patients, 4024 could be matched to at least 1 nonsurgical patient (Figure 1). Figure 1. Flow diagram for identification of eligible patients with T2DM and no history of microvascular disease who had bariatric surgery. BMI= body mass index; EHR= electronic health record; HbA1c= hemoglobin A1c; T2DM= type 2 diabetes mellitus. * Adults aged 2079 y who had a primary (first observed) bariatric surgical procedure between 1 January 2005 and 31 December 2011. Bariatric procedures were identified using a combination of bariatric registries; review of medical records; International Classification of Diseases, Ninth Revision, codes (43.89, 44.31, 44.38, 44.39, 44.68, 44.69, and 44.95); and Current Procedural Terminology codes (43633, 43644, 43645, 43659, 43770, 43775, 43842, 43843, 43844, 43845, 43846, and 43847). We excluded patients with any of the following: <1 y of continuous enrollment; history of gastrointestinal surgery for cancer; pregnancy in the year before surgery; gestational diabetes as the sole diabetes diagnosis; preexisting neuropathy, nephropathy, or retinopathy; metformin as the sole indicator of possible T2DM (no other T2DM medications, laboratory values, or diagnoses); or maximum BMI <35 kg/m2. Patients may have >1 type of missing data (BMI, creatinine concentration, or HbA1c level) or have no follow-up time after surgery. 98 surgical patients could not be matched to a nonsurgical patient on site, age, BMI, insulin use, HbA1c level, and sex. Appendix Table 1. Comparison of Characteristics of Surgical Patients Who Were Missing Values for BMI, HbA1c, or Serum Creatinine at Baseline Versus Those Who Had Complete Information on These Values at Baseline Matched Nonsurgical Participants For each surgical patient, we identified up to 3 matched nonsurgical control participants via a 2-step process. First, among all patients with diabetes and at least 1 BMI measurement of 35 kg/m2 or greater who did not have bariatric surgery during the study (n= 320345), we identified a pool of potential control participants who were enrolled at the time of the surgery; satisfied the study inclusion criteria; and matched the surgical patient on the basis of study site, sex, age (10 years), BMI (5 kg/m2), HbA1c level (1.5 percentage points), and insulin use. Second, for each control participant in the pool, we calculated Mahalanobis distance from the surgical patient on the basis of age, BMI, HbA1c level, diabetes duration, and total days of health care use in the 7 to 24 months before the date of surgery (21). Finally, up to 3 control participants were selected on the basis of the shortest Mahalanobis distance. Throughout, nonsurgical patients could be used as a control for only 1 surgical patient (matching without replacement). The Appendix gives additional details on the process we used to establish the matched cohort. Outcome and Censoring Definitions The primary outcome measure was time to incident microvascular disease (composite of the first occurrence of retinopathy, neuropathy, or nephropathy). Retinopathy was defined by the first appearance of either ICD-9 code 362.0x (diabetic retinopathy) or both a retinal surgical procedure code or other retinopathic code (Appendix Table 2) and ICD-9 code 250.5x (diabetes with ophthalmic manifestations). Neuropathy was identified on the basis of the first occurrence of ICD-9 code 250.6 (diabetes with neurologic manifestations) or 357.2 (polyneuropathy in diabetes). Finally, nephropathy was defined by at least 2 measures of estimated glomerular filtration rate less than 60 mL/min/1.73 m2 separated by at least 90 days without any intervening values of 60 mL/min/1.73 m2 or greater. We approximated estimated

The Authors Respond: A Multisite Study of Long-Term Remission and Relapse of Type 2 Diabetes Mellitus Following Gastric Bypass

We thank Drs. Hussain and El-Hasani for their thoughtful comments and critique of our manuscript. We appreciate their endorsement of our finding that patients with surrogate markers for more severe type 2 diabetes and worse beta cell function, i.e., those with longer duration of disease, insulin use, and poor preoperative glycemic control, are least likely to experience remission and are at higher risk for relapse of their diabetes after gastric bypass surgery [1]. Their first point about the standardization of technique is interesting, but unfortunately does not take into consideration a key feature of our study. Specifically, we conducted a study to examine the effectiveness of gastric bypass on diabetes remission and relapse in diverse health care settings involving numerous surgical centers and dozens of surgeons. Our study is to date the largest multicenter cohort with long-term follow-up available. Thus, our findings are more generalizable and representative of typical gastric bypass outcomes as practiced in large, integrated health care systems in California and Minnesota than findings from a single, albeit highly standardized, academic medical center. Furthermore, although it is an appealing concept, we are unaware of any high-quality data that link long-term type 2 diabetes remission to technical modifications in gastric bypass, and we encourage the authors to provide citations in support of this concept. Their second comment refers to “failed cases” in reference to the 35 % of gastric bypass patients in our study that experienced a relapse of their type 2 diabetes following an initial remission. We find the term “failure” unhelpful in this context and did not use it in our manuscript. Indeed, we believe that patients who experience even a transient remission of diabetes are likely to have substantial long-term health benefits, and we are currently undertaking a longerterm follow-up study of our cohort to examine the impact of gastric bypass surgery on incident microand macrovascular disease events. Finally, while it was not the main aim of our study, we did not find that weight regain was strongly linked to diabetes relapse. Instead, we found that the aforementioned preoperative surrogate markers of diabetes severity and poor beta cell function were most predictive of relapse. Their third point regarding our rates of disenrollment was previously discussed as a limitation in the original manuscript, and while we cannot test this hypothesis, we do not believe that patients lost to follow-up due to change in health plan insurance would have superior outcomes to those with continued health care in an integrated system. Their fourth point takes issue with our estimation of the remission rate and reflects a misunderstanding of our methods, results, and current diabetes treatment practices. First, we used the latest recommended definitions for remission and relapse of type 2 diabetes [2], and the observed rates of remission are in line with other recently published D. E. Arterburn (*) :A. Bogart :M. K. Theis Group Health Research Institute, Seattle, WA, USA e-mail: arterburn.d@ghc.org