IBSR Newsletter - Spring 2003

Spring 2003

IBSR Newsletter

Spring 2003 Volume 18, number 1

Mortality in Obesity Surgery

Edward E. Mason MD, Ph.D.

When the NBSR began in 1986, one of the goals was to collect enough data regarding different operations so that we could determine if any of them had survivorship advantages. We know that 30-day operative mortality for all operations has decreased by an order of magnitude since the beginning of obesity surgery 50 years ago. But, the IBSR operative mortality has risen in the last four years to 0.3% from 0.2% during the previous four years (unpublished data).

The IBSR performed a preliminary study of 2,363 men operated upon during the 13 years beginning 1-1-86 and ending 12-31-98.¹ There were 107 deaths found by the National Death Index (NDI). Table 1 shows when deaths occurred and how many were reported to IBSR before the NDI search. The IBSR knew about all but one of the deaths that occurred during the first 30-days and six of eight at six months. Beyond that, follow-up was inadequate for further study of the original IBSR data. A regression analysis of these data showed that operative age, weight, BMI, history of heart disease and history of kidney disease were significant predictors of 30-day postoperative mortality. Another 12 variables were not significant, including whether the operation was a simple restriction or a bypass procedure.

Deaths Following Surgery	IBSR	NDI
Within 30 days	16	17
One to six months	6	8
One year (+ 6 months)	4	9
Two to five years (+ 6 months)	11	32
Six to ten years (+ 6 months)	7	32
Greater than 10 years	0	9
TOTAL	44	107

Table 1: IBSR 2000 NDI search results for male deaths.

The NDI just completed a sixteen-year search from 1986-2001 on an IBSR subset of data collected for operations starting 1-1-86 to 12-31-1999. Of the 18,877 patient records examined over 8,000 partial matches occurred but were coded 0 (meaning not a probable death). However, approximately 400 records had the NDI status code of 1 (probable death). The analysis of these results is under way currently. We wanted members to begin thinking about what might be found and offer you the opportunity to participate further in the analysis of the data that were contributed by you. As explained in the Kathleen Renquist's article on page 2 of this newsletter, we are prohibited from providing identifiable death information to IBSR contributors.

The IBSR has information for 32,434 patients in 'Merge 32', but we submitted data from only 18,877. The difference in number is because only patients with complete Social Security Number (SSN) and full names from US surgeons were submitted. Other identifiers are gender, birth date, and last follow-up. The NDI is conservative in their matching. A probability of a match is calculated on what and how many data elements matched between our records and death certificate information to calculate status code 1. Even when the NDI status code is 1, there are a few instances in which the patient's recorded death was before the operation for obesity or last follow-up. This requires that we review each possible match and make a decision to accept or reject it as a death. For NDI status code 0 the matches are partial with different combinations of data elements matching. One example is a last name change through marriage. If all of the variables match except for last name, we accept that the patient has died. The NDI believes that they are able to find 95% of the patients that have died. Comparison of IBSR and NDI 30-day deaths makes the study worthwhile in that it provides an audit of the accuracy of the data that we rely on for reporting relative operative risk for different types of operations.

IBSR has devoted much time and effort in extending operative complications and mortality beyond discharge to an arbitrary 30-days following operation.² One of the major causes of death is pulmonary embolism and it is also one that remains in force during that entire time of the operation and beyond. One explanation could be the toxicity of free fatty acids, which activate the Hageman factor.³ There is another lethal complication that continues for life for patients with bypass operations, and that is obstruction of the biliopancreatic limb.

These are long awaited and exciting days at the IBSR. Thank you for your efforts in making the studies possible. Everyone should benefit from the added data regarding such an important variable.

1. Renquist KE, Mason EE, Neuhaus J, Zimmerman B, IBSR Contributors. Obtaining mortality data using the National Death Index. Obesity Surgery 10:111, 2000. (Abstract) Presented at the ASBS Allied Health Session, Memphis TN.

2. Mason EE, Renquist KE, Zimmerman B, IBSR Members. Age, BMI, and gender determine 30-day complications. Obesity Surgery 12: 215, 2002. (Abstract) Presented at the ASBS Convention; Las Vegas NV.

3. Mason EE, Gordy DD, Chernigoy FA, et al. Fatty acid toxicity. Surg Gynecol Obstet 133:992-998, 1971.

Obtaining Mortality Data and the National Death Index

By Kathleen Renquist, BS

Comparison of mortality data following surgical treatment of obesity has been a research goal of the IBSR since the NBSR first began in 1986. Do IBSR surgeons submit sufficient data within 30 days of the operation to study 30-day complication and mortality rates? We know that follow-up past 18 months is insufficient based on follow-up rate calculations. It is for these two reasons (verification and increased follow-up) that more complete data, or finding resources where it may be obtained, is very important.

The National Death Index (NDI) is a repository of death certificate information from every state in United States of America. It is maintained by the National Center for Health Statistics, Public Health Service, U.S. Department of Health and Human Services. An organization can gain approval to submit data to the NDI through a formal application process. The expense of performing a search matching the requester's data in NDI format can be formidable for large data sets.

NDI assurances must be met prior to approval for a search. Any identifying information can only be used for statistical purposes in medical or health research. Two, information will be used only for the study or project described in the approved NDI application. This means the IBSR can not provide identifiable death information to contributing surgeons if we find someone who died that is not in their submitted data set.

In 2000, the IBSR submitted 2,373 male patient records at the cost of $5,117.48. Important information was obtained from this male subset, but the total number of deaths (107) was insufficient to perform a survival analysis. We did find the number of deaths reported to the IBSR and the matched NDI search was consistent within 30 days of surgery. The three most frequent causes of death cardiac arrest, pulmonary embolism and obesity were similar to IBSR data. This pilot study suggested it would be possible to test a hypothesis with regard to the effect of initial BMI, age, gender and operative category upon survivorship in a larger data set.

In April 2003, an IBSR data subset of both males and females was submitted. The cost of searching death records for all 18,877 patients from 1986 to 2001 was $41,471.05. At the writing of this article, we are reviewing the NDI matches in preparation for further analyses.

IBSR 1.0 f Feature: "Mortality Lists"

By Kathleen Renquist, BS

One of the most frequent information requests from the IBSR is mortality rate. The pooled IBSR reports list death information according to days past the operation date with deaths within 30 days of the operation the central focus. Each user has the option to examine the completeness of this information in their data set.

The IBSR user can examine death information in their database in two ways. One is to use the Primary Field List for Mortality Information. This list provides the number of recorded deaths in the data set, patient idnumber, date of death, age of patient at death, weight and BMI (if weight is available), and cause of death according to the ICD.9.CM selected for entry when the death record was created. You may gain access to this information via Reports from the menu bar, Lists, Primary Field Lists, and then Mortality Information.

A second method for obtaining mortality information is to create the Mortality Report. This is done by using Reports from the menu bar, Lists, and then Mortality Data. An example of the summary of mortality information provide in the software is shown below. Both a mortality rate for patients who have had a primary bariatric procedure and those who have come to your office for a revision of a bariatric procedure performed by other surgeons are listed.

==================================================================

Operation Mortality Data

Today's date: 05-05-2003

Primary Operation Mortality Data:

The number of operations = 200

The number of deaths = 2

Mortality rate = (2)/(200) = 0.1%

The number alive = 198

Percentage of total = (198)/(200) = 99.9%

Primary Reoperation Mortality Data:

The number of reoperations = 20

The number of deaths = 1

Mortality rate = (1)/(20) = 5.0%

The number alive = 19

Percentage of total = (19)/(20) = 95.0%

=================================================================

first name last name date Cause (1) Cause (2) Cause (3)

John Smith 05/04/2002 Pulmonary embolism Cardiac arrest Respiratory Failure

=================================================================

In order for death information to appear on this report it must be entered under mortality information in the individual patient record. Do not record death information in the "other complications" of the operative record or "additional information" in follow-up as it will not be stored properly in the data set for reporting purposes. Be as complete as possible as the information recorded should provide the most accurate picture regarding the cause of a patient's death. Please examine your data sets for completeness of mortality information.

The IBSR Newsletter is published on a quarterly basis by the International Bariatric Surgery Registry. The IBSR takes due care to accurately report information from sources believed reliable, but cannot assume liability for any information published. Errors, when discovered, will be promptly corrected.

SPRING 2003 IBSR Newsletter Volume 18, Number 1

Within 30 days