Advances in the Prevention and Control of HAIs

Using Claims Data to Perform Surveillance for Surgical Site Infection: The Devil Is in the Details

Table of Contents

Katelin B. Nickel, Anna E. Wallace, David K. Warren, Daniel Mines, Margaret A. Olsen

Abstract

Increasingly, investigators and regulatory agencies are using billing claims data to identify healthcare-associated infections (HAIs) for quality improvement initiatives. We use hernia repair as an example to investigate methodological strategies for surgical site infection (SSI) surveillance and the effect of variation in coding of anatomical surgical site by provider and facility on SSI incidence. Using commercial insurer claims data, we conducted a retrospective cohort study of enrollees aged 6 months to 64 years with ICD-9-CM procedure or CPT-4 codes from facility and/or provider claims for umbilical, inguinal/femoral, and incisional/ventral hernia repairs, between 2004 and 2010. SSIs occurring within 90 days after hernia repair and before a subsequent surgery were identified by ICD-9-CM diagnosis codes. Supporting evidence for operations included UB-92 revenue codes and CPT-4 codes for pathology, mesh, and administration of anesthesia. A total of 181,811 hernia repair procedures were initially identified based on distinct procedure dates more than 7 days apart. The number of distinct procedures was reduced to 140,632 after removing procedures with no supportive evidence for operation, operations in medically complicated patients, complicated hernia operations, operations coded for more than one hernia site or unclassified, and operations performed at the time of an SSI. The incidence of SSIs was compared according to the stringency of identification of the hernia site. When agreement between the provider and the facility classification was required to define the hernia site, the incidence rates of SSI after incisional/ventral, inguinal/femoral, and umbilical hernia repairs were 4.11 percent (715/17,390), 0.45 percent (352/77,666), and 1.16 percent (288/24,917), respectively. By contrast, the incidence rates after hernia repair at the three sites were 3.29 percent (199/6,041), 0.53 percent (56/10,573), and 1.41 percent (57/4,045) for procedures with discordant coding and facility/provider-only information. Use of administrative data to identify SSI requires a thorough methodological approach to accurately identify and characterize surgical procedures, particularly when surgical factors are important risk factors for infection.

Introduction

Using claims data to determine the incidence of surgical site infection (SSI) after surgical procedures requires accurate identification of both infections and the surgical procedures. Many of the studies describing the use of administrative or claims data to perform SSI surveillance have focused only on the accuracy of International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM) diagnosis codes to identify the infections, with varying results, depending on the surgical procedures studied and the diagnosis codes used to indicate infection.1–8

Few investigators have validated the accuracy of ICD-9-CM procedure and Current Procedural Terminology, 4th edition (CPT-4®) codes to identify and characterize surgical procedures. This information is critical to accurately determine the denominator of at-risk procedures in the calculation of SSI incidence.9–14 The relative lack of studies validating the coding of surgical procedures may be due to the perception that procedures are more accurately coded than diagnoses.15

The overall goal of our study was to determine the incidence of SSI among procedures performed in inpatient hospital and ambulatory surgical centers (both hospital-based and free-standing centers), using claims data from a large U.S. private insurer. We developed a rigorous algorithm to identify procedures in order to ensure the accuracy of the calculated incidence of infection. We use the identification of unique hernia repair procedures as a complicated example, since these procedures can occur repeatedly within an individual; and classification of the procedures by site of repair is important because the incidence of SSI likely varies by anatomic site.16–19 In this study, it was important to characterize the anatomic site of hernia repair, since our ultimate goal was to compare SSI rates according to the location where surgery was performed. We assumed that this location would vary, based on the anatomic site of the hernia. We expected that incisional hernia repair would more likely be performed in an inpatient hospital setting than would inguinal, femoral, or umbilical hernia repairs. Therefore, it was important to develop methods to accurately characterize the anatomic site of the hernia in order to compare SSI rates within hernia types across the three types of facilities.

Methods

Data Source

We conducted a retrospective cohort study using the HealthCore Integrated Research Database™ (HIRD), which contains longitudinal claims data from 13 WellPoint-owned Blue Cross and/or Blue Shield health plans located in the Southeast, Mid-Atlantic, Eastern, Central, and Western regions of the United States. Data in the HIRD include all fully adjudicated claims submitted for reimbursement from providers, facilities, and outpatient pharmacies and are linked to health plan enrollment information.

Fully insured members enrolled in a health plan that included, at a minimum, medical coverage of hospital and physician services were eligible for selection into the study cohort. Members with an ICD-9-CM diagnosis code or prescription claim that indicated HIV-positive status at any time during the study period were not eligible for entry into the cohort because HIV status could not be disclosed due to privacy considerations. Members who were likely to have incomplete data also were not eligible. These included members enrolled in a plan with hospital coverage only, since up to 60 percent of SSIs are identified and managed in the ambulatory setting;20 members enrolled in a plan that reimburses providers through capitated, rather than fee-for-service payment, since utilization data are likely incomplete;21 and members enrolled in multiple plans at the time of the surgery of interest. Operations on members whose insurance coverage ended on the day of hernia repair were excluded because ascertainment of subsequent SSI would not be possible. Medical claims were restricted to paid claims.

The research data for this study contained up to five ICD-9-CM diagnosis codes and five ICD-9-CM procedure codes per claim. Inpatient facility claims also included Uniform Billing (UB-92) revenue and Healthcare Common Procedure Coding System (HCPCS) codes, while ambulatory facility and provider claims included CPT-4 and HCPCS codes. Service dates were available for lines with CPT-4 codes, but exact procedure dates were not available from inpatient facility claims, which included only the first service date (interpreted as the inpatient admission date) and the last service date (i.e., the discharge date).

Hernia Repair Patient Population

We initially identified hernia repair surgeries among members eligible for cohort entry (n=171,140 members, each with one or more hernia procedures) and aged 6 months to 64 years, from January 1, 2004 through December 31, 2010, using ICD-9-CM and CPT-4 procedure codes from inpatient and ambulatory facilities (other than home health agencies) and provider claims for incisional/ventral, inguinal/femoral, and umbilical hernia repairs (Table 1). The hernia repair patient population was further refined by excluding operations likely to have erroneous claims for hernia repair, complicated procedures and operations in patients considered medically complicated, and procedures in which the surgery date and/or classification of the hernia site could not be determined from the available information in the claims (Figure 1).

Table 1. ICD-9-CM procedure and CPT-4 codes used to identify hernia repair

Operative Category Laparoscopic Open
ICD-9-CM CPT-4 ICD-9-CM CPT-4
Incisional / ventral 53.62, 53.63, 54.21* + (53.51, 53.61, 53.59, 53.69) 49654–49657 53.51, 53.61, 53.59, 53.69 49560, 49561, 49565, 49566
Inguinal / femoral 17.11–17.13, 17.21–17.24, 54.21* + (53.00–53.05, 53.10–53.17, 53.21, 53.29, 53.31, 53.39) 49650, 49651 53.00–53.05, 53.10–53.17, 53.21, 53.29, 53.31, 53.39 49500, 49501, 49505, 49507, 49520, 49521, 49525, 49550, 49553, 49555, 49557
Umbilical 53.42, 53.43, 54.21* + (53.41, 53.49) 49652, 49653 53.41, 53.49 49580, 49582, 49585, 49587

*Required that 54.21 be on the same claim as the open hernia ICD-9-CM procedure code.

Figure 1. Claims data algorithm for classification of hernia repair

Figure 1 contains a flowchart detailing the claims data algorithm for classification of hernia repair. After grouping by 7-day windows, the starting population consisted of 181,811 hernia repair procedures. We then excluded 6,500 procedures with no supporting evidence for operation, 6,392 procedures in medically-complicated patients, 21,113 complicated procedures, 6,748 procedures that were either not able to be classified or were from more than one hernia site, and 426 procedures with a pre-existing surgical site infection. Of the remaining 140,632 procedures, we compared 119,973 procedures with matching provider and facility hernia classification to 20,659 hernia repair procedures that had discordant classification information between the provider and facility, or were provider-only or facility-only.

Problem Claims

We created an algorithm to identify "problem claims." Problem claims were those that contained CPT-4, HCPCS, or UB-92 revenue codes truncated to four digits and populated in the fields reserved for ICD-9-CM procedure codes. This error apparently occurred during processing of certain types of non-inpatient facility claims (authors' unpublished observations). Claims in which a hernia procedure code was present only on one line on a single claim, with no other claims on the same date, were also classified as problem claims.

Complicated Patients and Operations

The primary aim of this study was to estimate the risk of SSI among hernia procedures, depending on the surgical facility type. For this reason, we excluded hernia repair procedures performed in medically complicated patients, who would be very unlikely to undergo surgery in an ambulatory setting. These patients would have unique risk factors that would put them at higher risk of SSI compared with other patients. The medically complicated patient category included persons coded for cancer or chemotherapy, or with end-stage renal disease, from 30 days before to 7 days after the hernia procedure. Hernia repairs in persons with septicemia between 7 days before to 1 day after the hernia procedure date were excluded because those individuals had preexisting systemic infection. We also excluded hernia repairs in persons with motor vehicle accidents, abdominal compartment syndrome, or gunshot wounds coded on the same line as the hernia repair (Table 2).

In addition, we excluded complicated hernia repair procedures performed at the time of or after another surgical procedure during the same admission. The rationale for this exclusion is that these procedures would be complex, and if an SSI developed, it would not be possible to determine the attributable surgery. Additional surgical procedures were identified using CPT-4 and ICD-9-CM procedure codes for an operative procedure on the National Healthcare Safety Network (NHSN) list of procedures for SSI surveillance, including hernia sites not included in this study (epigastric, lumbar, spigelian, and omphalocele). 22 We also used CPT-4 codes for a variety of abdominal surgeries that are not part of the NHSN list, or are included in the NHSN list only as ICD-9-CM procedure codes, to exclude additional abdominal procedures performed before the hernia during the same admission or on the same day as the hernia repair (Table 2). We expanded the identification of other abdominal procedures through the use of CPT-4 codes for two reasons: first, because the majority of other surgical procedures performed before or at the time of hernia repair involved the abdomen, and second, to define more accurately the date of the other surgical procedure, since the research database did not contain procedure dates for ICD-9-CM procedure codes. We did not want to exclude procedures that occurred after the hernia repair because they may have been performed as a result of a complication of the hernia operation. The ability to more precisely date the additional procedures using CPT-4 codes with their associated service date was thus very important.

Table 2. Codes used for surgery exclusion and as evidence for surgery

  CPT-4 or HCPCS Codes ICD-9-CM Procedure Codes UB-92 Revenue Codes ICD-9-CM Diagnosis Codes
Codes Used for Hernia Repair Exclusion
Abdominal surgery codes 43279–43425, 43600–43659, 43770–43999, 44010–44130, 44133, 44136–44139, 44155–44158, 44211–44238, 44345, 44346, 44602–44900, 44950–45170, 45395–45505, 45540–45825, 46260–46288, 47010, 47100–47142, 47370, 47400–47460, 47560, 47561, 47570, 47579, 47700–47999, 48105–48160, 48520–48548, 49203–49220, 49255–49329      
Non–study hernia codes (epigastric, lumbar, spigelian, omphalocele) 49540, 49570, 49572, 49590, 49600, 49605, 49606, 49610, 49611      
Cancer       140.0–172.9, 174.0–209.36, 209.70–209.79
Chemotherapy 96400–96549, J9000–J9999, Q0083–Q0085 99.25 0331, 0332, 0335 V58.11, V58.12, V66.2, V67.2
End-stage renal disease       585.6, V45.1, V45.11, V45.12, V56.0, V56.1, V56.2, V56.8
Septicemia       038.0–0.38.9, 790.7
Motor vehicle accidents       E810.0–E825.9
Abdominal compartment syndrome       729.73
Gunshot wounds       E922.0–E922.3, E922.8, E922.9, E928.7, E965.0–E965.4, E97.0, E979.4, E985.0–E985.4
Codes Used as Additional Evidence for Hernia Repair Surgery
Anesthesia 00750, 00752, 00830, 00832, 00834, 00840, 00860      
Pathology 88302      
Surgical mesh (individual codes) C1781, S2077, 49568      
Surgery-related revenue codes     201, 360, 361, 369, 370, 379, 490, 499, 963, 964, 975  

Finally, we also excluded the remaining hernia repairs performed on or after day 3 of an inpatient hospital stay (with admission date considered day 1) because these patients would not have had the opportunity for surgery in an ambulatory facility.

Characteristics of the Surgery

Establishing the date of surgery and classifying the hernia site were fundamental to the analysis of the study. Operations lacking these surgical characteristics based on the claims were excluded.

Establishing the Surgery Date and Use of Supporting Evidence for Surgery

Hernia repair dates within 7 days were collapsed into a single surgery date because of the potential inaccuracy in dates, particularly on provider claims.23 In addition to the procedure codes in Table 1, CPT-4 code 49659 (unlisted laparoscopy procedure, hernioplasty, herniorrhaphy, herniotomy) coded by a provider was included as supplemental information to refine the surgery date when none of the CPT-4 codes in Table 1 were used by a provider.

When there was more than one date within 7 days that coded for hernia repair, we compared facility and provider surgery dates and incorporated supplemental claims evidence from other unique providers. We used these supplemental claims, which included anesthesia and pathology, along with claims for surgical mesh, to determine the most likely surgery date. For hernia repairs coded only by a provider or by a facility, we required additional evidence that an operation actually took place, including anesthesia, pathology, or surgical mesh codes or a surgery-related UB-92 revenue code (Table 2). If no additional evidence existed for facility-only or provider-only claims, the surgeries were excluded. Because procedures coded using ICD-9-CM procedure codes lacked an exact service date, the hernia repair date from the inpatient facility stay defaulted to the admission date. We therefore prioritized the service date from the provider (when available) to assign the surgery date, together with the supplemental information from other unique providers.

After determining the most likely hernia surgery date within the 7-day window, further steps to define the exact date of operation were performed based on the hospital length of stay and the number of days between multiple hernia repair dates. This allowed us to further consolidate dates that appeared to refer to the same hernia procedure. We consolidated multiple hernia surgery dates within inpatient admissions of more than 4 days, since it was unlikely that more than one hernia repair procedure took place during a single admission.

Classification of Hernia Site

We classified the hernia repairs as incisional/ventral, inguinal/femoral, and/or umbilical, based on the ICD-9-CM procedure and CPT-4 codes from the provider and facility (Table 1). If claims were available from both a provider and a facility, and the classification of hernia site differed, we prioritized the site and approach (i.e., laparoscopic vs. open) coded by the provider because the information coded by the performing surgeon should be more accurate. We excluded hernia procedures when the provider coded for a laparoscopic hernia repair with an unlisted site (CPT-4 code 49659) and the facility coded for an open hernia repair of a specified site, since both the hernia site and the technique were unclear. If the final hernia categorization involved more than one site (i.e., if the provider coded for more than one site or the surgery was coded only by a facility that coded for more than one hernia site), we excluded that surgery since it would be a more complex procedure and could not be classified to a single site.

Identification of Primary Outcome—Surgical Site Infection

SSIs first recorded from 2 to 90 days after eligible surgeries were identified using ICD-9-CM diagnosis codes from inpatient and ambulatory facilities and provider claims. We excluded claims with locations that were not consistent with a provider diagnosis (e.g., laboratory, patient's home), as well as claims with CPT-4 codes for pathology services (88104–88399), because the diagnosis codes on those lines may have been indicators of tentative or "rule-out" diagnoses.

The codes used to identify SSI included the standard postoperative wound infection codes (998.5, 998.51, 998.59, 996.69) plus codes for peritonitis (567.2–567.29, 567.9) and retroperitoneal infection (567.3–567.39). In accordance with the NHSN definition of SSI,22 a diagnosis code of cellulitis of the trunk (682.2) or unspecified site (682.9) on the same claim as a CPT-4 code for incision and drainage (10060, 10061, 10180, 11005, 11008, 49020, 49021, 49040, 49041, 49060, 49061) was considered evidence of SSI. Because the ICD-9-CM diagnosis code 682.9 refers to cellulitis and abscess at an unspecified site, it was used as an indicator of SSI only if it met one of the following criteria: if it was on the same claim line as an abdomen-specific CPT-4 procedure code (11005, 11008, 49020, 49021, 49040, 49041, 49060, 49061), or if it was coded by the surgeon who performed the hernia repair on the same claim as a non-abdomen-specific CPT-4 code for incision and drainage (10060, 10061, 10180).

The date of onset of SSI was defined according to the timing and location of diagnosis. For SSI newly coded by an inpatient facility during the original operative admission, we assigned the date of SSI to the discharge date if the difference between the discharge and admission date was greater than or equal to 2 days. For SSI diagnosed during a subsequent inpatient admission, the date of SSI onset was assumed to be the date of hospital admission. For SSI diagnosed initially in an ambulatory setting, the onset date was defined as the first service date with an ICD-9-CM diagnosis code for SSI.

ICD-9-CM diagnosis codes for SSI from 30 days before to 1 day after surgery were considered preexisting infection. Infections coded on the day of or the day after surgery were considered preexisting because the development and diagnosis of SSI due to hernia repair requires a minimum of 1 day; therefore, infections coded within 1 day of surgery most likely represent prevalent infection. We excluded hernia repairs if a preexisting SSI was identified.

The observation period for development of SSI was through 90 days after surgery, with earlier censoring for the end of insurance enrollment, a subsequent hernia repair, or another abdominal surgery. For subsequent surgeries, we censored 1 day after the subsequent surgery to be consistent with our algorithm that SSI coded the day after a surgical procedure is preexisting and can be attributed to the previous surgery. Non-abdomen-specific ICD-9-CM diagnosis codes for infection (e.g., 998.59) were not classified as SSI if they were first coded after a subsequent non-abdominal NHSN surgery within 90 days.

The incidence of SSI within 90 days of surgery by hernia site was compared using the chi-square test. All data management and statistical analyses were performed using SAS v9.2 (SAS Institute Inc., Cary, NC).

Results

A total of 181,811 hernia repair procedures were initially identified during the 7-year period examined in this study, based on distinct procedure dates more than 7 days apart. The number of distinct procedures was reduced to 140,632 (termed the "complete surgical set," go to Figure 1) among 135,907 patients, after removing procedures with no supporting evidence for operation (n=6,500), procedures in medically complicated patients (n=6,392), complicated hernia procedures (n=21,113), hernia repairs coded for more than one site or unclassified (n=6,748), and hernia repairs performed at the time of an existing SSI (n=426).

Of the 140,632 hernia repair procedures, a total of 119,973 (85.3 percent) procedures in 116,572 patients had an exact facility and provider match for both the hernia site and the approach; this group formed a subset for further analysis (Figure 1). Of the remaining 20,659 hernia repair procedures, 6,398 (4.5 percent of the complete surgical set) had discordant information from the provider and the facility regarding the hernia site and/or use of laparoscopy; in these cases the provider information was used to characterize the procedure. A total of 6,404 (4.6 percent of the complete surgical set) hernia procedures were defined by information from the facility only, and 7,857 (5.6 percent of the complete surgical set) were based on information from the provider(s) only. Of the complete surgical set, 23,431 (16.7 percent) procedures were classified as incisional/ventral hernia repairs, 88,239 (62.7 percent) as inguinal/femoral hernia repairs, and 28,962 (20.6 percent) as umbilical hernia repairs. Among the subset of 119,973 hernia repair procedures with matched facility-provider classification, 17,390 (14.5 percent) were classified as incisional/ventral hernia repairs, 77,666 (64.7 percent) as inguinal/femoral repairs, and 24,917 (20.8 percent) as umbilical hernia repairs.

A total of 1,667 incident SSIs were identified after the 140,632 hernia surgeries, for an overall SSI rate of 1.19 percent. The incidence of SSI was 3.90 percent (914/23,431) after incisional/ventral, 0.46 percent (408/88,239) after inguinal/femoral, and 1.19 percent (345/28,962) after umbilical hernia repairs. The incidence of SSI within 90 days was significantly lower in unmatched incisional/ventral hernia procedures compared to the incidence following procedures in which provider and facility coding agreed on the site and approach (3.29 percent SSI in unmatched incisional/ventral hernia repair vs. 4.11 percent SSI in incisional/ventral hernia repair with concordant facility-provider coding, p=0.005). For both inguinal/femoral and umbilical hernia repairs, the calculated incidence of SSI was higher in unmatched procedures compared to procedures with concordant coding of the site and approach of repair, although the differences were not statistically significant (Table 3).

Table 3. Comparison of SSI rates by classification of hernia repair

Hernia Site Matched Provider / Facility Classification (n=119,973) Discordant Classification or Provider / Facility Only (n=20,659) p
SSI n(%) Total procedures SSI n(%) Total procedures
Incisional / ventral 715 (4.11) 17,390 199 (3.29) 6,041 0.005
Inguinal / femoral 352 (0.45) 77,666 56 (0.53) 10,573 0.277
Umbilical 288 (1.16) 24,917 57 (1.41) 4,045 0.168

 

Discussion

In many studies using claims data, ICD-9-CM procedure codes alone, or a combination of ICD-9-CM procedure and CPT-4 codes from both facility and provider claims data, are used to identify surgical procedures in inpatient facilities. In the majority of these studies, detailed algorithms to identify surgical procedures are not provided. Detailed methods to identify procedures are very important in performing surveillance for healthcare-associated infections (HAIs) because attribution of infection to the surgery requires that the surgery actually occurred and that it took place before the infection. Therefore, rigorous methods to identify surgical procedures and to determine the most likely procedure date are essential if claims data are to be used reliably for HAI surveillance.

Using the example of hernia repair, we developed systematic strategies to increase the likelihood that a surgery actually took place; to refine the surgery date, taking into account the possibility of repeated surgeries within an individual; and to characterize the anatomic site of surgery. These steps are important components of SSI surveillance after all surgical procedures, in particular, ensuring that a procedure took place (to avoid overpopulating the denominator with non-procedures) and refining the surgery date (to ensure that only incident infections are identified). The steps we performed to refine the surgery date and to increase the likelihood that a surgical procedure took place reduced the number of hernia repair procedures by 3.6 percent (6,500/181,811). This reduction removed duplicate or non-surgeries from the denominator. The number of hernia procedures was reduced by an additional 14.7 percent when we required complete agreement between the provider and the facility regarding the hernia surgery site and approach.

We determined some of the methodological steps necessary to clean the data from experience and the literature, whereas others were developed after working with the commercial insurer data. Determining the correct denominator for surgical procedures is straightforward when detailed clinical information is available (e.g., operative log), but it can be more challenging when claims data are used. For example, procedure codes can sometimes be included on preoperative or postoperative medical claims, complicating identification of the surgery date. The accuracy of ICD-9-CM procedure codes to identify specific surgical procedures is less than 100 percent, and thus inpatient facility dates may not always correspond to a surgery date for the procedure of interest. The lack of procedure dates from inpatient facilities made it necessary to develop an algorithm to calculate the most likely surgical date, particularly when the surgery took place during a hospitalization lasting more than a few days. Determining the correct surgical date is also important to discriminate between preexisting infections and SSIs attributable to the procedure.

By requiring agreement in coding of anatomic site and approach from the provider and the facility, the likelihood of accuracy of classification will likely be higher, since the coding by providers and facilities is performed independently. The requirement for concordant information regarding the classification of procedures should therefore lead to more accurate estimates of the incidence of SSI. Although the NHSN does not stratify hernia SSI rates based on anatomic location, we suspected, based on the published literature,17,24 that incisional/ventral hernia surgeries would have a higher incidence of SSI compared to inguinal hernia, which was confirmed. We found that the incidence of SSI following incisional/ventral hernia repair was higher when we required provider and facility coding agreement, and the incidence of SSI for umbilical and inguinal/femoral hernia repair decreased when we required provider and facility concordance. We suspect that some of the discordant incisional/ventral hernia repairs involved other anatomic sites (e.g., inguinal), and thus the SSI incidence was lower for these imprecisely coded procedures. In the case of the discordant inguinal/femoral and umbilical hernia repairs, it is possible that some of the imprecisely coded procedures were actually incisional/ventral repairs, and thus the SSI incidence was higher for the discordant procedures than for the umbilical procedures with matched facility and provider coding.

By definition, the use of claims data for SSI surveillance involves secondary analysis of the data collected for billing purposes. Therefore, some data elements that are important for SSI surveillance, such as procedure dates, may be less important for reimbursement and, as a result, may be less accurate or coded with less detail. There is also the potential for misclassification of diagnoses and likely undercoding of SSIs, particularly minor infections during the 90-day global surgical reimbursement period for providers.25 Thus our calculations for the incidence of SSI after the three types of hernia repair are likely underestimates of the true infection rates after these procedures.

Some authors have concluded that billing and claims data cannot be reliably used for HAI surveillance, including SSIs.4,26 We believe that use of careful and methodical strategies to deal with the inconsistencies in claims data and algorithms tailored to specific surgical procedures, such as we have described for hernia repair, will improve the accuracy of ICD-9-CM diagnosis codes to identify true SSIs. Claims data cannot be approached as simple data that can easily be mined to identify infections. Rather, the individual datasets must be carefully analyzed to determine the extent of missing information (e.g., not all ICD-9-CM diagnosis and procedure codes submitted for reimbursement may be provided by the data source) and how this missing information affects the determination of type of procedure, procedure date, and infection. Comparison of facility and provider surgical coding can provide insight into strategies to identify surgical procedures more accurately. Knowledge of the billing and claims processes is important to understand how and when SSIs are coded. Clinical knowledge is also important when specific features of the surgical procedure are potentially associated with risk of SSI, as we determined for the anatomic site of hernia repair. We believe that the use of claims data for surveillance of SSIs requires the development of algorithms based on clinical knowledge of the surgical procedures and infection, together with a thorough analysis of the claims data, to expose problems and issues that may impair accurate identification of the procedures and SSIs. Only by subjecting the data to careful scrutiny can methodological strategies be developed to overcome the inconsistencies of claims data.

Conclusions

HAI surveillance based on claims data requires methodological approaches to identify both the infections and the denominators as accurately as possible. In the case of SSI, accurate identification of surgical procedures is essential to avoid overpopulation of the denominator with implausible events. We developed a systematic algorithm using complete claims data containing both ICD-9-CM procedure and CPT-4 codes from facilities and providers to exclude procedures that were unlikely to have been performed. We believe that more thorough approaches such as these are essential to determine accurate rates of SSI using complex claims data.

Acknowledgments

This project was funded under contract no. 5R01HS019713 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services. The findings and conclusions in this document are those of the authors, who are responsible for its content, and do not necessarily represent the views of AHRQ. No statement in this report should be construed as an official position of AHRQ or of the U.S. Department of Health and Human Services. We thank Cherie Hill for database and computer management support.

Authors' Affiliations

Division of Infectious Diseases, Department of Medicine (KBN, DKW, MAO), Division of Public Health Sciences, Department of Surgery (MAO), Washington University School of Medicine, St. Louis, MO. HealthCore, Inc., Wilmington, DE (AEW, DM).

Address correspondence to: Margaret A. Olsen, Ph.D., MPH, Division of Infectious Diseases, Campus Box 8051, Washington University, 660 S. Euclid Ave., St. Louis, MO 63110; Email: molsen@dom.wustl.edu.

References

1. Yokoe DS, Noskin GA, Cunningham SM, et al. Enhanced identification of postoperative infections among inpatients. Emerg Inf Dis 2004 Nov;10(11):1924-30. PMID: 15550201.

2. Bolon MK, Hooper D, Stevenson KB, et al. Improved surveillance for surgical site infections after orthopedic implantation procedures: extending applications for automated data. Clin Infect Dis 2009 May 1;48(9):1223-9. PMID: 19335165.

3. Olsen MA, Fraser VJ. Use of diagnosis codes and/or wound culture results for surveillance of surgical site infection after mastectomy and breast reconstruction. Infect Control Hosp Epidemiol 2010 May;31(5):544-7. PMID: 20334508.

4. West J, Khan Y, Murray DM, et al. Assessing specific secondary ICD-9-CM codes as potential predictors of surgical site infections. Am J Infect Control 2010 Nov;38:701-5. PMID: 20579771.

5. Calderwood MS, Ma A, Khan YM, et al. Use of Medicare diagnosis and procedure codes to improve detection of surgical site infections following hip arthroplasty, knee arthroplasty, and vascular surgery. Infect Control Hosp Epidemiol 2012 Jan;33(1):40-9. PMID: 22173521.

6. Hollenbeak CS, Boltz MM, Nikkel LE, et al. Electronic measures of surgical site infection: implications for estimating risks and costs. Infect Control Hosp Epidemiol 2011 Aug;32(8):784-90. PMID: 21768762.

7. Lawson EH, Louie R, Zingmond DS, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg 2012 Dec;256(6):973-81. PMID: 23095667.

8. Yokoe DS, Khan Y, Olsen MA, et al. Enhanced surgical site infection surveillance following hysterectomy, vascular, and colorectal surgery. Infect Control Hosp Epidemiol 2012 Aug;33(8):768-73. PMID: 22759543.

9. Wang MC, Laud PW, Macias M, et al. Strengths and limitations of International Classification of Disease Ninth Revision Clinical Modification codes in defining cervical spine surgery. Spine (Phila Pa 1976) 2011 Jan 1;36(1):E38-44. PMID: 20975624.

10. Wang MC, Laud PW, Macias M, et al. Utility of a combined current procedural terminology and International Classification of Diseases, Ninth Revision, Clinical Modification code algorithm in classifying cervical spine surgery for degenerative changes. Spine (Phila Pa 1976) 2011 Oct 15;36(22):1843-8. PMID: 21245779.

11. Bozic KJ, Chiu VW, Takemoto SK, et al. The validity of using administrative claims data in total joint arthroplasty outcomes research. J Arthroplasty 2010 Sep;25(6 Suppl):58-61. PMID: 20570479.

12. Meguerditchian AN, Stewart A, Roistacher J, et al. Claims data linked to hospital registry data enhance evaluation of the quality of care of breast cancer. J Surg Oncol 2010 June 1;101(7):593-9. PMID: 20461766.

13. Singh JA, Kundukulam JA, Bhandari M. A systematic review of validated methods for identifying orthopedic implant removal and revision using administrative data. Pharmacoepidemiol Drug Saf 2012 Jan;21 Suppl 1:265-73. PMID: 22262616.

14. Singh JA, Ayub S. Accuracy of VA databases for diagnoses of knee replacement and hip replacement. Osteoarthritis Cartilage 2010 Dec;18(12):1639-42. PMID: 20950694.

15. McCarthy EP, Iezzoni LI, Davis RB, et al. Does clinical evidence support ICD-9-CM diagnosis coding of complications? Med Care 2000 Aug;38(8):868-76. PMID: 10929998.

16. McGreevy JM, Goodney PP, Birkmeyer CM, et al. A prospective study comparing the complication rates between laparoscopic and open ventral hernia repairs. Surg Endosc 2003 Nov;17(11):1778-80. PMID: 12958679.

17. Neumayer L, Giobbie-Hurder A, Jonasson O, et al. Open mesh versus laparoscopic mesh repair of inguinal hernia. N Engl J Med 2004 Apr 29;350(18):1819-27. PMID: 15107485.

18. Mason RJ, Moazzez A, Sohn HJ, et al. Laparoscopic versus open anterior abdominal wall hernia repair: 30-day morbidity and mortality using the ACS-NSQIP database. Ann Surg 2011 Oct;254(4):641-52. PMID: 21881493.

19. Dalenbäck J, Andersson C, Ribokas D, et al. Long-term follow-up after elective adult umbilical hernia repair: low recurrence rates also after non-mesh repairs. Hernia 2012 Sep 13. PMID: 22971796.

20. Sands K, Vineyard G, Platt R. Surgical site infections occurring after hospital discharge. J Infect Dis 1996 Apr;173(4):963-70. PMID: 8603978.

21. Warren JL, Feuer E, Potosky AL, et al. Use of Medicare hospital and physician data to assess breast cancer incidence. Med Care 1999 May;37(5):445-56. PMID: 10335747.

22. Centers for Disease Control and Prevention. National Healthcare Safety Network (NHSN) Procedure-Associated (PA) Module: Surgical Site Infection (SSI) event. Available at: www.cdc.gov/nhsn/PDFs/pscManual/9pscSSIcurrent.pdf (PDF File, 233 KB).

23. Li X, King C, deGara C, et al. Validation of colorectal cancer surgery data from administrative data sources. BMC Med Res Methodol 2012 Jul 11;12:97. PMID: 22784239.

24. Choi JJ, Palaniappa NC, Dallas KB, et al. Use of mesh during ventral hernia repair in clean-contaminated and contaminated cases: outcomes of 33,832 cases. Ann Surg 2012 Jan;255(1):176-80. PMID: 21677561.

25. Centers for Medicare & Medicaid Services. Global Surgery Fact Sheet. Available at: www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/GloballSurgery-ICN907166.pdf (PDF File, 83 KB).

26. Julian KG, Brumbach AM, Chicora MK, et al. First year of mandatory reporting of healthcare-associated infections, Pennsylvania: an infection control-chart abstractor collaboration. Infect Control Hosp Epidemiol 2006 Sep;27(9):926-30. PMID: 16941317.

Return to Contents

Page last reviewed June 2014
Page originally created June 2014
Internet Citation: Using Claims Data to Perform Surveillance for Surgical Site Infection: The Devil Is in the Details. Content last reviewed June 2014. Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/professionals/quality-patient-safety/patient-safety-resources/resources/advances-in-hai/hai-article13.html