Materials and Methods
A multiinstitution deidentified electronic health record database, EPM: Explore (Explorys Inc, Cleveland, OH), was utilized to perform a retrospective cohort study to assess the impact of pregnancy on ventral and incisional hernia recurrence. EPM: Explore is a commercially available analytics platform that collects and standardizes deidentified data from a variety of health information systems including inpatient and ambulatory electronic health records, billing systems, and laboratory and radiology systems.
The deidentified data from participating health care organizations are collected into a cloud-based data storage grid, which is updated at least once every 24 hours. Each participating health care organization has access to a secure, web-based application that allows for exploration and analysis of population-level data.
At the time of this study, EPM: Explore contained data on more than 50 million patients from 26 integrated health systems and 360 hospitals across the United States. Detailed information on the EPM: Explore platform, data standardization and mapping have been previously published. Studies conducted on the EPM: Explore database are considered exempt by our institutional review board.
All women aged 18–45 years with a history of a primary abdominal hernia repair, defined as either a ventral or incisional hernia, between the years 1999 and 2013 were identified. Given that various hernia subtypes have different baseline recurrence risks, women with other hernia subtypes, such as umbilical, femoral, inguinal, or hiatal, were excluded to prevent conflation or misattribution of risk between subtypes. The presence or absence of a singleton pregnancy following primary repair was elucidated from the database. To exclude spontaneous abortion and early pregnancy loss, a pregnancy event was considered to have occurred only if a women had record of a vaginal, operative vaginal, or cesarean delivery.
Notably, the EPM: Explore platform allows users to place temporality restrictions on and between data fields within a search. For example, to ensure that a pregnancy event occurred after a hernia repair, we restricted our search to vaginal, operative vaginal, or cesarean deliveries occurring at least 12 months from the date of surgery for primary hernia repair. Therefore, although search or exclusion by gestational age is not available within EPM: Explore, the platform provides a sufficient mechanism to accurately identify a cohort of women with a pregnancy resulting in delivery after a primary hernia repair. Women with pregnancy-related primary hernia repairs (occurring during pregnancy or the 6 week postpartum period) were excluded to eliminate women with concomitant exposure.
The presence or absence of a clinically significant hernia recurrence, defined as reoperation, within the first 5 years after primary repair was then abstracted. A 5 year time frame from primary hernia repair was chosen to capture a plausible window in which a pregnancy could have an impact on the risk of hernia recurrence and to reduce confounding by other chronic risk factors. Additional demographic and obstetric data were obtained along with information regarding the presence or absence of risk factors for hernia recurrence including diabetes, obesity (body mass index >30 kg/m 2 prior to hernia repair), tobacco abuse, and a wound complication at the time of initial hernia repair (infection or dehiscence).
Demographic and medical comorbidity data were abstracted from the time of the primary hernia repair. The rate of hernia recurrence was calculated for the entire cohort and for women with and without a history of a pregnancy after primary repair. The association between pregnancy and hernia recurrence was evaluated with a logistic regression, both unadjusted and adjusted for pertinent risk factors.
Risk factors that were significant in the univariable analysis ( P < .1) were maintained in the final multivariable model. Age was not incorporated in the multivariable model because EPM: Explore reports only categorical age data. Given that prior evidence has not demonstrated an impact of race on the risk of hernia recurrence and that race was essentially dichotomized (white, African American, or other), race was also eliminated from the final multivariate model.
Lastly, because diabetes and obesity are correlated, an interaction variable was added to the final multivariable model. To maintain Health Insurance Portability and Accountability Act–compliant statistical deidentification, EPM: Explore reports population counts rounded to the nearest 10 and does not report sample sizes less than 10.
To determine whether such rounding had an adverse impact on the results, calculations of hat matrix and Pregibon’s d-beta were performed to assess for undue influence and leverage, respectively. Categorical variables were assessed using a χ 2 or Fisher exact test as appropriate. Odds ratios with 95% confidence intervals are presented. Data analysis was conducted using STATA version 13.1 (Stata Corp, College Station, TX).
Results
A total of 11,020 women with a history of a hernia repair were identified, of whom 840 had a subsequent pregnancy. Baseline demographic and clinical information, stratified by history of pregnancy after hernia repair, is presented in Table 1 . Women with a pregnancy after hernia repair were younger, more likely to be insured by Medicaid, and more likely to have risk factors for hernia recurrence including obesity, tobacco abuse, and wound complication at the time of initial hernia repair.
| Characteristics | Pregnancy after repair (n = 840) | No pregnancy after repair (n = 10,180) | P value |
|---|---|---|---|
| Age, y | < .001 | ||
| 15–20 | 0 (0) | 50 (0.5) | |
| 20–29 | 150 (17.9) | 520 (5.1) | |
| 30–39 | 470 (56.0) | 2670 (26.2) | |
| 40–45 | 210 (25.0) | 6940 (68.2) | |
| Race | < .001 | ||
| White | 500 (59.5) | 7210 (70.8) | |
| African American | 300 (35.7) | 1960 (19.3) | |
| Other | 40 (4.8) | 1010 (9.9) | |
| Insurance status | |||
| Private | 520 (61.9) | 6150 (60.4) | < .001 |
| Medicaid | 310 (36.9) | 2200 (21.6) | |
| Self-pay/other | 10 (1.2) | 1730 (17.0) | |
| Obesity (BMI >30 kg/m 2 ) | 620 (73.8) | 6010 (59.3) | < .001 |
| Diabetes | 150 (17.9) | 1950 (19.2) | .35 |
| Wound complication | 60 (7.1) | 470 (4.6) | < .001 |
| Tobacco | 200 (23.8) | 1590 (15.6) | < .001 |
Overall, 915 women in the cohort had hernia recurrence (8.3%). A univariable analysis of risk factors for hernia recurrence is presented in Table 2 . As expected, all other traditional risk factors for hernia recurrence occurred more frequently among women experiencing a recurrence. In unadjusted analysis, pregnancy was associated with an increase in the risk of hernia recurrence (13.1% vs 7.1%, odds ratio, 1.96, 95% confidence interval, 1.60–2.42).
| Clinical factor | Hernia recurrence (n = 915) | No hernia recurrence (n = 10,105) | OR (95% CI) |
|---|---|---|---|
| Pregnancy | 120 (13.1) | 720 (7.1) | 1.96 (1.60–2.42) |
| BMI >30 kg/m 2 | 670 (80.9) | 5960 (59.0) | 2.31 (1.98–2.69) |
| Diabetes | 260 (28.4) | 1840 (18.2) | 2.05 (1.78–2.39) |
| Wound complication | 110 (12.0) | 420 (4.2) | 3.76 (3.05–4.64) |
| Tobacco | 200 (21.9) | 1590 (15.7) | 1.73 (1.47–2.03) |
Multivariable logistic regression was then performed to assess the impact of pregnancy on hernia recurrence with adjustment for confounding factors. Covariates in the model included pregnancy, obesity (body mass index >30 kg/m 2 ), diabetes, tobacco abuse, and wound dehiscence at the time of initial hernia repair. The interaction between diabetes and obesity was not significant in the final multivariate model ( Figure ).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
