- Email: [email protected]
Risk factors for urinary tract infection in the postpartum period
Risk factors for urinary tract infection in the postpartum period Margot A. Schwartz, MD, MPH,a, c Chia C. Wang, MD, MS, a, c Linda O. Eckert, MD,b and Cathy W. Critchlow, PhDc Seattle, Washington OBJECTIVE: We sought to examine risk factors for urinary tract infection in postpartum women. STUDY DESIGN: Subjects (n = 931) with maternal urinary tract infections and control subjects (n = 1862) were identified by using a linked Washington State birth certificate and Birth Events Records Database for the years 1987-1993; stratified analysis was performed by using Mantel-Haenszel procedures. RESULTS: Increased risk for postpartum urinary tract infection was associated with black, Native American, or Hispanic race-ethnicity (odds ratio, 1.30; 95% confidence interval, 1.03-1.64) and unmarried status (odds ratio, 1.33; 95% confidence interval, 1.11-1.58). Cesarean delivery (odds ratio, 2.70; 95% confidence interval, 2.27-3.20) and tocolysis (odds ratio, 3.30; 95% confidence interval, 2.15-5.06) also contributed to maternal risk of acquiring a urinary tract infection. Maternal risk factors included renal disease (adjusted odds ratio, 3.89; 95% confidence interval, 1.80-8.41) and preeclampsia-eclampsia (adjusted odds ratio, 3.21; 95% confidence interval, 2.36-4.38). Among women undergoing vaginal delivery, renal disease (odds ratio, 5.47; 95% confidence interval, 2.04-14.64) and abruptio placentae (odds ratio, 5.02; 95% confidence interval, 1.84-13.64) were risk factors. Length of hospital stay was significantly associated with urinary tract infection. CONCLUSION: Maternal medical conditions and procedures that predispose to urinary tract infections are those that also are associated with urethral catheterization. In addition, maternal urinary tract infections may contribute significantly to duration of postpartum hospital stay. (Am J Obstet Gynecol 1999;181:547-53.)
Key words: Catheterization, epidemiology, length of stay, puerperal infections, urinary tract infection
Approximately 20% of all women reportedly seek treatment for a urinary tract infection at some point during their lifetime.1, 2 Infections of the female urethra and bladder may occur spontaneously or as a result of sexual activity or instrumentation of the urogenital tract. In pregnancy additional physiologic and structural changes contribute to the susceptibility to urinary tract infection. Moreover, in those deliveries involving instrumentation, catheterization, or surgical procedures, the risk of acquiring a urinary tract infection may be additionally increased. Although risk factors for asymptomatic bacteriuria3-6 and urinary tract infection7 during pregnancy have been well characterized, fewer data are available regarding risk factors for postpartum urinary tract infection.
From the Departments of Medicinea and Obstetrics and Gynecology,b School of Medicine, and the Department of Epidemiology, School of Public Health and Community Medicine,c University of Washington. Supported in part by a National Institute of Allergy and Infectious Diseases STD/AIDS Research Training Grant (National Institute of Health Grant T32 AI 07140) and by an STD/HIV Prevention Fellowship sponsored by the Association of Teachers of Preventive Medicine and the Centers for Disease Control and Prevention. Received for publication June 25, 1998; revised May 3, 1999; accepted May 21, 1999. Reprints not available from the authors. Copyright © 1999 by Mosby, Inc. 0002-9378/99 $8.00 + 0 6/1/100254
Identification of such risk factors could result in earlier detection and treatment of infection, with potential reductions in maternal morbidity, hospital stay, and medical costs associated with postpartum care. Antibiotic prophylaxis for women at high risk could be considered. Our aim was to identify socioeconomic factors, medical conditions, and obstetric procedures contributing to the risk of acquiring a urinary tract infection in the postpartum period by performing a case-control study using Washington State birth certificate data linked to hospitalization records for the years 1987 to 1993. Methods We conducted a population-based case-control study using Washington State birth certificate data linked to the Birth Events Records Database for the years 19871993. This latter database contains hospital discharge information for both mother and child for birth hospitalizations in Washington state and is created by matching the Comprehensive Hospital Abstract Reporting System database to birth records by name, date of birth, hospital, sex, zip code, and other factors. Study subjects were those women with a diagnosis of urinary tract infection (designated by the International Classification of Diseases, Ninth Revision [ICD-9] code 599.0) listed in the hospital discharge record (n = 931), which includes the 5 major diagnoses and 3 major proce547
548 Schwartz et al
September 1999 Am J Obstet Gynecol
Table I. Demographic characteristics of women with and without urinary tract infection in Washington State, 1987-1993 Control subjects (n = 1862) Characteristic Mother’s age 15-19 y 20-29 y ≥30 y Ethnicity Black, Hispanic, Native American Other nonwhite White Mother’s education level1 prior birth 0-1 prior births
Case subjects (n = 931) No.
%
Odds ratio* and 95% confidence interval
No.
%
232 1074 556
12.5 57.7 29.9
127 531 273
13.6 57.0 29.3
1.11 (0.85-1.46) 1.01 (0.84-1.21) Reference
220 84 1520
12.1 4.6 83.3
140 35 745
15.2 3.8 81.0
1.30 (1.03-1.64) 0.85 (0.56-1.30) Reference
198 176 1488
10.6 9.5 79.9
102 69 760
11.0 7.4 81.6
1.01 (0.78-1.31) 0.77 (0.57-1.04) Reference
467 401 48 1373
25.1 23.6 4.2 73.7
286 213 35 689
30.8 25.2 6.3 74.0
1.33 (1.11-1.58) 1.10 (0.91-1.33) 1.52 (0.98-2.38) 1.01 (0.85-1.21)
1227 101
92.4 7.6
634 38
94.4 5.7
1.37 (0.94-2.02) Reference
444 1384
24.3 75.7
204 721
22.1 78.0
0.88 (0.73-1.07) Reference
*Crude odds ratios estimate the risk of urinary tract infection among women with the characteristic compared with women in the reference category.
dures on the face sheet of the hospital discharge papers. Because our focus was postpartum urinary tract infection, we excluded 76 cases of pyelonephritis, which most likely reflected antepartum urinary tract infection. Control subjects (n = 1862) were randomly selected in a 2:1 control subject/study subject ratio from the pool of women in the database without a discharge diagnosis of urinary tract infection. To avoid misclassification, we excluded potential control subjects who may have had urinary tract infections that were misclassified as other conditions. These diagnoses included major puerperal infection (ICD-9 code 670); inflammatory diseases of the ovary, fallopian tubes, or peritoneum (ICD-9 code 614); inflammatory diseases of the cervix, vagina, or vulva (ICD-9 code 616); other sexually transmitted diseases (ICD-9 codes 090-099.9); pyelonephritis (ICD-9 code 590); and infections of the genitourinary tract in pregnancy (ICD-9 code 646.6). To validate the diagnosis of urinary tract infection in the database and to assess for information regarding catheterization and peripartum antibiotic use, we reviewed charts from 37 women with urinary tract infections who were delivered of their neonates at the University of Washington Medical Center. We examined the effect of hospital procedures on the prevalence of urinary tract infections, such as fetal monitoring, induction or stimulation of labor or both, tocolysis, cesarean delivery, vacuum extraction, manual extraction, and forceps extraction, as potential risk factors.
Internal and external fetal monitoring were not distinguishable in our database. Also, induction and stimulation of labor were combined for the purpose of this analysis because, although labor induction may be a longer procedure, both procedures frequently involve use of epidural anesthesia and bladder catheterization. We posited that these procedures may increase the risk of urinary tract infection because they might alter urinary tract function, result in an increased number of vaginal examinations with or without instrumentation, thereby potentially contaminating the urethra with vaginal or rectal flora, or cause urethral injury. In addition, we considered epidural anesthesia as a potential risk factor; however, this variable was incompletely reported in our database. Demographic factors of interest included race, age, marital status, educational level, income, location of birth setting, parity, smoking status, and alcohol use. Black, Hispanic, and Native American race were combined into 1 race-ethnicity variable because of the small numbers in each group. In addition, the maternal conditions diabetes, renal disease (including acute and chronic renal disease and nephrolithiasis, but excluding infection), obesity, anemia, cephalopelvic disproportion, abruptio placentae, placenta previa, and preeclampsia or eclampsia (a combined diagnosis) were examined as risk factors for postpartum urinary tract infection. These factors reflect maternal health and immune status, as well as the degree of complication associated with delivery. In addition, the association between postpartum urinary
Schwartz et al 549
Volume 181, Number 3 Am J Obstet Gynecol
Table II. Obstetric procedures and complications among women with and without urinary tract infection in Washington State, 1987-1993 Control subjects (n = 1862)
Case subjects (n = 931)
Condition
No.
%
No.
%
Cesarean delivery Fetal monitoring† Labor induction‡ Outlet forceps Tocolysis Manual extraction Vacuum extraction Perineal laceration Episiotomy
367 963 445 171 33 268 150 493 528
20.7 51.7 23.9 9.7 2.7 14.4 8.1 26.5 28.4
374 481 251 81 51 125 55 169 229
41.3 51.7 27.0 9.1 8.4 13.4 5.9 18.2 24.6
Crude odds ratio and 95% confidence interval* 2.70 (2.27-3.20) 1.00 (0.85-1.17) 1.18 (0.98-1.41) 0.93 (0.70-1.22) 3.30 (2.15-5.06) 0.92 (0.73-1.16) 0.72 (0.52-0.99) 0.62 (0.51-0.75) 0.82 (0.69-0.99)
*Crude odds ratios estimate the risk of urinary tract infection among women with as compared with those without the stated condition. †Includes both internal and external monitoring. ‡Includes both induction and stimulation of labor.
tract infection and delivery of a low-birth-weight (<2500 g) infant was explored. The risk of acquiring a postpartum urinary tract infection associated with hospital procedures, maternal medical conditions, and demographic characteristics was estimated by odds ratios derived by using Mantel-Haenszel procedures.8 Statistical analyses were carried out by using SAS (SAS Institute, Cary, NC). Presence of confounding was assessed by comparing crude and adjusted odds ratios obtained from stratified analyses. A variable was considered to be a confounder if it was associated with both the risk factor and urinary tract infection and if the odds ratio adjusted for the confounder differed from the crude odds ratio by ≥10%. We specifically evaluated whether cesarean delivery modified associations between urinary tract infection and hospital procedures, maternal diagnoses, and demographic characteristics. Results Women who acquired urinary tract infection were similar to those who did not acquire uninary tract infection with respect to age, educational level, smoking, alcohol use, income, location of birth, and parity (Table I). Women of black, Hispanic, or Native American race were slightly more likely than white women to have urinary tract infections (odds ratio, 1.30; 95% confidence interval, 1.03-1.64); however, small numbers of women in each race group precluded examining individual race groups as a predictor for the presence of urinary tract infection. Unmarried women also demonstrated a slightly greater risk than married women (odds ratio, 1.33; 95% confidence interval, 1.11-1.58). The frequency of procedures performed differed between patients and control subjects (Table II). Women who were delivered of their infants by cesarean were nearly 3 times more likely to acquire a postpartum urinary tract infection than women who were not (odds
ratio, 2.70; 95% confidence interval, 2.27-3.20), as were women who received tocolytic agents (odds ratio, 3.30; 95% confidence interval, 2.15-5.06). Cesarean delivery did not significantly confound the relationship between tocolysis and postpartum urinary tract infection. Vacuum extraction, perineal laceration, and episiotomy were more common in control subjects than in women with urinary tract infections and initially appeared to be associated with a decreased risk of urinary tract infection (Table II). This association no longer remained after cesarean delivery was controlled for (data not shown). Women with urinary tract infections and control subjects were similar with respect to prevalence of gestational or preexisting diabetes, obesity, and anemia but differed with respect to preexisting renal disease (odds ratio, 3.77; 95% confidence interval, 1.69-8.41), preeclampsia-eclampsia (odds ratio, 3.62; 95% confidence interval, 2.72-4.84), and pregnancies or deliveries complicated by abruptio placentae (odds ratio, 2.17; 95% confidence interval, 1.06-4.44), placenta previa (odds ratio, 3.67; 95% confidence interval, 1.77-7.59), cephalopelvic disproportion (odds ratio, 1.58; 95% confidence interval, 1.09-2.27), or breech presentation (odds ratio, 1.84; 95% confidence interval, 1.32-2.56; Table III). However, few women in both the urinary tract infection and control groups had renal disease, abruptio placentae, or placenta previa. After we adjusted for cesarean section, both renal disease (odds ratio, 3.89; 95% confidence interval, 1.80-8.41) and preeclampsia-eclampsia (odds ratio, 3.21; 95% confidence interval, 2.36-4.38) remained as significant maternal risk factors for postpartum urinary tract infection (Table IV). When stratified by method of delivery, renal disease was a significant risk factor for postpartum urinary tract infection among women who were delivered vaginally (odds ratio, 5.47; 95% confidence interval, 2.04-14.64) but not among
550 Schwartz et al
September 1999 Am J Obstet Gynecol
Table III. Maternal medical conditions and pregnancy complications among women with and without urinary tract infections in Washington State, 1987-1993 Control subjects (n = 1862) Condition Diabetes† Obesity‡ Anemia Renal disease§ Preeclampsiall Cephalopelvic disproportion Abruptio placentae Placenta previa Breech delivery
Case subjects (n = 931)
No.
%
No.
%
Crude odds ratio and 95% confidence interval*
48 60 32 8 72 69 14 10 78
2.6 14.9 2.7 0.5 4.2 4.0 0.8 0.6 4.5
36 30 19 15 120 53 15 18 69
3.9 14.7 3.2 1.8 13.8 6.1 1.7 2.1 8.0
1.52 (0.98-2.35) 0.99 (0.62-1.59) 1.20 (0.68-2.14) 3.77 (1.69-8.41) 3.62 (2.72-4.84) 1.58 (1.09-2.27) 2.17 (1.06-4.44) 3.67 (1.77-7.59) 1.84 (1.32-2.56)
*Crude odds ratios estimate the risk of urinary tract infection among women with as compared with those without the stated condition. †Includes linked information for gestational and other diabetes. ‡Defined as prepregnancy weight of ≥180 pounds. §Includes linked data for acute or chronic renal disease and nephrolithiasis but excludes infection. llIncludes both eclampsia and preeclampsia.
women who were delivered by cesarean (odds ratio, 1.43; 95% confidence interval, 0.24-8.59). The number of women who had both renal disease and a cesarean delivery, however, was small. Labor induction or stimulation was of borderline significance as a risk factor for postpartum urinary tract infection both before and after we adjusted for cesarean delivery. When stratified by method of delivery, women who were delivered vaginally and underwent labor induction or stimulation had a slightly increased risk of urinary tract infection (odds ratio, 1.35; 95% confidence interval, 1.09-1.68), although, among women who were delivered of their infants by cesarean, induction of labor was not associated with urinary tract infection (odds ratio, 0.99; 95% confidence interval, 0.70-1.41). Abruptio placentae was not a significant risk factor for postpartum urinary tract infection after we adjusted for cesarean section (odds ratio, 1.78; 95% confidence interval, 0.93-3.42). However, among women who underwent vaginal delivery, abruptio placentae was associated with an increased risk of urinary tract infection (odds ratio, 5.02; 95% confidence interval, 1.84-13.64). Next, we further explored the observed associations between occurrence of postpartum urinary tract infection with race-ethnicity and marital status. The association between black, Hispanic, or Native American race and urinary tract infection was no longer statistically significant after adjustment by preeclampsia-eclampsia (adjusted odds ratio, 1.25; 95% confidence interval, 0.981.61). The association of unmarried status with postpartum urinary tract infection, however, was not changed after adjustment by either urban birthplace (adjusted odds ratio, 1.35; 95% confidence interval ,1.101.66) or mother’s age (adjusted odds ratio 1.34; 95% confidence interval, 1.11-1.61).
Low-birth-weight infants were more common among women with urinary tract infections than among control subjects (15.2% vs 5.4%; odds ratio, 3.11; 95% confidence interval, 2.40-4.03). Of note, low birth weight was also more common among women who gave birth in an urban setting (9.5% vs 4.3%; odds ratio, 2.32; 95% confidence interval, 1.03-5.20), among unmarried women (12.4% vs 7.2%; OR = 1.81, 95% confidence interval, 1.38-2.37), and among women of black, Hispanic, or Native American race (15.3% vs 7.5%; odds ratio, 2.24; 95% confidence interval, 1.59-3.14). Thirty-seven women with urinary tract infection who gave birth at the University of Washington Medical Center were randomly selected for chart review. Of these, 32 had urinary tract infection diagnosed by urine culture, and the remainder had a diagnosis by urinalysis alone. Finally, we examined the association between length of hospital stay and maternal diagnosis of urinary tract infection. Because the risk of acquiring a urinary tract infection is known to be strongly associated with bladder catheterization and because we did not have specific data as to whether catheterization had been performed, we examined the association of urinary tract infections with length of hospital stay separately among women presumably at high, as opposed to low, catheterization risk. The group at high risk for catheterization included those women undergoing tocolysis or cesarean delivery or those with preeclampsia-eclampsia; those without these conditions were considered to be at low risk for catheterization. Length of stay was significantly associated with urinary tract infection among women both at low (P value for test of trend < .001) and at high (P value for test of trend < .001) likelihood of having had bladder catheterization performed (Table V).
Schwartz et al 551
Volume 181, Number 3 Am J Obstet Gynecol
Table IV. Maternal medical conditions, procedures, and pregnancy complications in Washington State, 1987-1993: Risk estimates for urinary tract infection adjusted for cesarean delivery Control subjects (n = 2128) Condition Renal disease* Cesarean delivery No cesarean delivery Preeclampsia† Cesarean delivery No cesarean delivery Cephalopelvic disproportion Cesarean delivery No cesarean delivery Breech presentation Cesarean delivery No cesarean delivery Placenta previa Cesarean delivery No cesarean delivery Abruptio placentae Cesarean delivery No cesarean delivery Labor induction‡ Cesarean delivery No cesarean delivery
Case subjects (n = 1064) No.
%
Adjusted odds ratio and 95% confidence interval
No.
%
2/328 6/1359
0.6 0.4
3/346 12/507
0.9 2.4
3.89 (1.80-8.41)
18/331 53/1359
5.4 3.9
65/349 52/514
18.6 10.1
3.21 (2.36-4.38)
68/342 1/1383
19.9 0.1
52/347 1/514
15.0 0.2
0.73 (0.49-1.08)
64/347 14/1383
18.4 1.0
64/350 5/514
18.3 1.0
0.99 (0.69-1.41)
9/342 1/1383
2.6 0.1
16/347 2/514
4.6 0.4
2.00 (0.94-4.28)
8/342 6/1383
2.3 0.4
4/347 11/514
1.2 2.1
1.78 (0.93-3.42)
78/367 361/1405
21.2 25.7
79/374 169/531
21.1 31.8
1.24 (1.03-1.49)
*Renal disease combines data for acute or chronic renal disease or nephrolithiasis but excludes infection. †Includes both eclampsia and preeclampsia. ‡Includes both induction and stimulation of labor.
Comment This is a large case-control study that confirms the data of 2 prior studies9, 10 demonstrating cesarean delivery to be an important risk factor for postpartum urinary tract infection. These data also identify 2 additional procedures, tocolysis and induction of labor, as risk factors for urinary tract infection. Women who undergo these procedures often undergo urethral catheterization, which is estimated to result in urinary tract infection in 10% of patients.11, 12 Maternal renal disease, preeclampsiaeclampsia, and unmarried status were also risk factors for postpartum urinary tract infection. Additionally, length of hospital stay was also associated with risk of urinary tract infection. Identification of cesarean delivery, tocolysis, and induction of labor as urinary tract infection risk factors may underscore the prominence of bladder catheterization as a cause of nosocomial urinary tract infection. Unfortunately, the information in this database did not identify women who had received bladder catheterization. Even if catheterization data had been available, the catheterization risk for postpartum urinary tract infection may still have been difficult to separate from risk associated with tocolysis, cesarean delivery, or induction of labor because many of these women also would have had catheterization. It may be that the risk of urinary tract infection conferred by these procedures is actually mediated or caused by bladder catheterization. Prospective,
controlled studies as to the presence and length of continuous versus intermittent catheterization would be helpful. Prior studies have reported maternal risk factors for postcesarean infectious complications, including urinary tract infection. In a prospective study Parrott et al13 found obesity and low socioeconomic status to be risk factors for postcesarean urinary tract infection. Nielsen and Hokegard14 found obesity, anemia, duration of labor, and duration of ruptured membranes to be associated with an increased risk of postcesarean nosocomial infection. Unlike these investigators, we did not find anemia, obesity, diabetes, or income to be associated with occurrence of urinary tract infection. There may be several explanations for the greater than 3-fold risk of urinary tract infection in women with renal disease or preeclampsia. Both preeclampsia and renal disease are frequently associated with proteinuria, which may enhance a woman’s risk for acquiring a urinary tract infection.15 Renal disease could also signify anatomic abnormalities (eg, reflux nephropathy, scarring, or nephrolithiasis providing a nidus for infection) or loss of immunoglobulins in the urine contributing to a relatively immunosuppressed state. Additionally, women with either renal disease or preeclampsia may require bladder catheterization more frequently than control subjects. Many of the maternal conditions or complications of delivery frequently requiring cesarean delivery, such as
552 Schwartz et al
September 1999 Am J Obstet Gynecol
Table V. Maternal urinary tract infection and length of hospital stay in women in Washington State, 1987-1993, stratified by risk of bladder catheterization Control subjects (n = 1862)* Length of stay (d) Low catheterization risk‡ 1 2 3 4 ≥5 High catheterization risk§ ≤2 3 4 ≥5
Study subjects (n = 931)
No.
%
No.
%
Odds ratio† and 95% confidence interval
466 597 174 31 16
36.4 46.4 13.6 2.4 1.2
101 199 113 48 96
18.2 35.7 20.3 8.6 17.2
Reference 1.54 (1.17-2.03) 3.00 (2.15-4.18) 7.14 (4.21-12.15) 27.68 (15.16-51.19)
20 129 124 47
6.3 40.3 38.8 14.7
6 83 120 165
1.6 22.2 32.1 44.1
Reference 2.14 (0.77-6.25) 3.23 (1.17-9.32) 11.70 (4.13-34.77)
*Two hundred fifty-eight control subjects were missing data on length of hospital stay. †Odds ratios compare risk of urinary tract infection with that in women in the specified reference category. ‡Includes women who did not have preeclampsia and did not undergo tocolysis or cesarean delivery. §Includes women who had preeclampsia or underwent tocolysis or cesarean delivery; 1 and 2 days were combined because no patients at high risk for catheterization stayed only 1 day.
cephalopelvic disproportion, breech delivery, and placenta previa, were associated with urinary tract infection by univariate analysis. However, when adjusted for cesarean delivery, these procedures were not found to be significant risk factors for urinary tract infection. This suggests that the risk thought to be associated with these procedures was presumably caused by catheterization performed as a result of cesarean delivery. Although cesarean delivery potentially confounded the relationship between abruptio placentae and urinary tract infection, abruptio placentae was a risk factor for postpartum urinary tract infection in women who underwent vaginal delivery. Many of these women may have undergone catheterization as a result of being at bed rest or for monitoring maternal volume status and urine output. The association between delivery of a low-birth-weight infant and postpartum urinary tract infection, which has not been previously reported, may be attributed to a complex interplay of demographic factors. However, we did not find that race or marital status confounded the association between low birth weight and the presence of a urinary tract infection. Asymptomatic bacteriuria of pregnancy, a known risk factor for low birth weight and prematurity,6, 16-19 is found more often in women of low socioeconomic status.3, 4 Furthermore, the association between unmarried status and urinary tract infections, also observed by Hooton et al,20 may reflect behavioral factors, such as frequency of sexual intercourse or exposure to new partners. The relationship between length of hospital stay and maternal diagnosis of urinary tract infection has important implications and deserves further study. A retrospective cohort study of women who underwent cesarean delivery found that urinary tract infection significantly prolonged hospital stay (7.2 days compared with 5.7 days, P < .001).21 Although a postpartum urinary tract infec-
tion may result in a longer hospital stay, it may also be that the nosocomial risks associated with longer hospital stays, such as use of an indwelling bladder catheter, increase the incidence of postpartum urinary tract infection. Additionally, a longer hospital stay allows a greater opportunity for a clinician to make a urinary tract infection diagnosis. We attempted to adjust for factors that reflect a complicated delivery, including cesarean, tocolysis, and low birth weight, and found that the association remained. However, this case-control analysis does not allow a cause-and-effect determination between urinary tract infection and length of hospital stay. Other limitations of our study result from the use of birth certificate data and hospital discharge data, which were not specifically designed or originally collected for research purposes. The hospital discharge record, for example, lists only 5 diagnoses and 3 procedures. Parrish et al22 examined the accuracy of birth certificate and hospital discharge data in Washington State and detected a 94% match between the birth certificate and the mother’s and infant’s hospital discharge abstracts. In the same study the ability of this data source to correctly identify obstetric procedures, including cesarean delivery, was found to be >90% for most types of delivery. However, it was also found that obstetric procedures not related to reimbursement, such as induction of labor and episiotomy, were not reliably reported in the database. Next, criteria for the diagnosis of urinary tract infections may have varied among our study subjects. Our chart review at one urban tertiary medical center in Washington State indicates that most women given a diagnosis of urinary tract infection in this population had the diagnosis made on the basis of urinalysis and culture. If less stringent criteria were used to make a urinary tract infection diagnosis, this misclassification would produce risk estimates lower than the true ones. On the other
Volume 181, Number 3 Am J Obstet Gynecol
hand, some women with urinary tract infection, such as those who did not have a urinalysis or culture or those occurring after hospital discharge, may have been missed. By excluding pyelonephritis from our case definition, other potential study subjects may have been missed as well. In addition, some subjects in our study may not have had postpartum urinary tract infection but rather antepartum asymptomatic bacteriuria or symptomatic urinary tract infection. These types of misclassification errors may also have resulted in underestimation of the true odds ratios. Despite the limitations, this study contributes important information. First, our findings demonstrate the risk of urinary tract infection associated with cesarean delivery. Second, this study identifies tocolysis as an additional risk factor. Although we could not directly examine bladder catheterization as a risk factor for nosocomial urinary tract infection, the procedures that we found to be risk factors often require bladder catheterization, thus indirectly highlighting its importance. Third, renal disease and preeclampsia were identified as additional risk factors that may also be associated with catheterization. Fourth, we report a strong association between postpartum urinary tract infection and length of hospital stay. These findings reinforce the practice of avoiding bladder catheterization when possible and further delineate medical conditions that might predispose a woman to postpartum urinary tract infection. Further research to confirm our findings might include information on prepartum and postpartum urinalyses and cultures, as well as data on bladder catheterization, epidural anesthesia, and reason for prolonged hospital stay. Attention to such factors could decrease maternal morbidity and cost in the immediate postpartum period. We thank Dr Pat Starzyk of the Washington State Center for Health Statistics for providing the data, Mr Bill O’Brien for assistance with data management, and Drs Beth Mueller, Steve Schwartz, and Walter Stamm for their advice and support. REFERENCES
1. Ronald AR, Pattullo ALS. The natural history of urinary tract infection in adults. Med Clin North Am 1991;75:299-312. 2. Weissenbacher ER, Reisenberger K. Uncomplicated urinary tract infections in pregnant and non-pregnant women. Curr Opin Obstet Gynecol 1993;5:513-6.
Schwartz et al 553
3. Turck M, Goffe BS, Petersdorf RG. Bacteriuria of pregnancy. N Engl J Med 1962;266:857-60. 4. Williams GL, Campbell H, Davies KJ. The influence of age, parity, and social class on the incidence of asymptomatic bacteriuria in pregnancy. J Obstet Gynaecol Br Commonw 1969;76:22939. 5. Layton R. Infection of the urinary tract in pregnancy: an investigation of a new routine in antenatal care. J Obstet Gynaecol Br Commonw 1964;71:927-33. 6. Savage WE, Hajj SN, Kass EH. Demographic and prognostic characteristics of bacteriuria in pregnancy. Medicine 1967;46:385-407. 7. Chng PK, Hall MH. Antenatal prediction of urinary tract infection in pregnancy. Br J Obstet Gynaecol 1982;89:8-11. 8. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22:719-48. 9. Buchholz N-P, Daly-Grandeau E, Huber-Buchholz M-M. Urological complications associated with caesarean section. Eur J Obstet Gynecol Biol 1994;56:161-3. 10. Leigh DA, Emmanuel FXS, Sedgwick J, Dean R. Post-operative urinary tract infection and wound infection in women undergoing caesarean section: a comparison of two study periods in 1985 and 1987. J Hosp Infect 1990;15:107-16. 11. Johnson JR, Roberts PL, Olsen RJ, Moyer KA, Stamm WE. Prevention of catheter-associated urinary tract infection with a silver oxide-coated urinary catheter: clinical and microbiologic correlates. J Infect Dis 1990;162:1145-50. 12. Stamm WE. Catheter-associated urinary tract infections: epidemiology, pathogenesis, and prevention. Am J Med 1991;91(Suppl 3B):65S-71. 13. Parrott T, Evans AJ, Lowes A, Dennis KJ. Infection following caesarean section. J Hosp Infect 1989;13:349-54. 14. Nielsen TF, Hokegard K-H. Postoperative cesarean section morbidity: a prospective study. Am J Obstet Gynecol 1983;146:911-6. 15. Cruikshank DP, Hays PM. Maternal physiology in pregnancy. In: Gabbe SG, Niebyl JR, Simpson JL, editors. Obstetrics: normal and problem pregnancy. New York: Churchill Livingstone; 1991. p. 125-46. 16. Kincaid-Smith P, Bullen M. Bacteriuria in pregnancy. Lancet 1965;1:395-9. 17. DeBaun M, Rowley D, Province M, Stockbauer JW, Cole FS. Selected antepartum medical complications and very-low-birthweight infants among black and white women. Am J Public Health 1994;84:1495-7. 18. Schieve LA, Handler A, Hershow R, Persky V, Davis F. Urinary tract infection during pregnancy: its association with maternal morbidity and perinatal outcome. Am J Public Health 1994;84:405-10. 19. Kass EH. Pyelonephritis and bacteriuria: a major problem in preventive medicine. Ann Intern Med 1962;56:46-53. 20. Hooton TM, Scholes D, Hughes JP, et al. A prospective study of risk factors for symptomatic urinary tract infection in young women. N Engl J Med 1996;335:468-74. 21. Henderson E, Love EJ. Incidence of hospital-acquired infections associated with caesarean section. J Hosp Infect 1995;29:245-55. 22. Parrish KM, Holt VL, Connell FA, Williams B, LoGerfo JP. Variations in the accuracy of obstetric procedures and diagnoses on birth records in Washington State, 1989. Am J Epidemiol 1993;138:119-27.
Key words: Catheterization, epidemiology, length of stay, puerperal infections, urinary tract infection
Approximately 20% of all women reportedly seek treatment for a urinary tract infection at some point during their lifetime.1, 2 Infections of the female urethra and bladder may occur spontaneously or as a result of sexual activity or instrumentation of the urogenital tract. In pregnancy additional physiologic and structural changes contribute to the susceptibility to urinary tract infection. Moreover, in those deliveries involving instrumentation, catheterization, or surgical procedures, the risk of acquiring a urinary tract infection may be additionally increased. Although risk factors for asymptomatic bacteriuria3-6 and urinary tract infection7 during pregnancy have been well characterized, fewer data are available regarding risk factors for postpartum urinary tract infection.
From the Departments of Medicinea and Obstetrics and Gynecology,b School of Medicine, and the Department of Epidemiology, School of Public Health and Community Medicine,c University of Washington. Supported in part by a National Institute of Allergy and Infectious Diseases STD/AIDS Research Training Grant (National Institute of Health Grant T32 AI 07140) and by an STD/HIV Prevention Fellowship sponsored by the Association of Teachers of Preventive Medicine and the Centers for Disease Control and Prevention. Received for publication June 25, 1998; revised May 3, 1999; accepted May 21, 1999. Reprints not available from the authors. Copyright © 1999 by Mosby, Inc. 0002-9378/99 $8.00 + 0 6/1/100254
Identification of such risk factors could result in earlier detection and treatment of infection, with potential reductions in maternal morbidity, hospital stay, and medical costs associated with postpartum care. Antibiotic prophylaxis for women at high risk could be considered. Our aim was to identify socioeconomic factors, medical conditions, and obstetric procedures contributing to the risk of acquiring a urinary tract infection in the postpartum period by performing a case-control study using Washington State birth certificate data linked to hospitalization records for the years 1987 to 1993. Methods We conducted a population-based case-control study using Washington State birth certificate data linked to the Birth Events Records Database for the years 19871993. This latter database contains hospital discharge information for both mother and child for birth hospitalizations in Washington state and is created by matching the Comprehensive Hospital Abstract Reporting System database to birth records by name, date of birth, hospital, sex, zip code, and other factors. Study subjects were those women with a diagnosis of urinary tract infection (designated by the International Classification of Diseases, Ninth Revision [ICD-9] code 599.0) listed in the hospital discharge record (n = 931), which includes the 5 major diagnoses and 3 major proce547
548 Schwartz et al
September 1999 Am J Obstet Gynecol
Table I. Demographic characteristics of women with and without urinary tract infection in Washington State, 1987-1993 Control subjects (n = 1862) Characteristic Mother’s age 15-19 y 20-29 y ≥30 y Ethnicity Black, Hispanic, Native American Other nonwhite White Mother’s education level
Case subjects (n = 931) No.
%
Odds ratio* and 95% confidence interval
No.
%
232 1074 556
12.5 57.7 29.9
127 531 273
13.6 57.0 29.3
1.11 (0.85-1.46) 1.01 (0.84-1.21) Reference
220 84 1520
12.1 4.6 83.3
140 35 745
15.2 3.8 81.0
1.30 (1.03-1.64) 0.85 (0.56-1.30) Reference
198 176 1488
10.6 9.5 79.9
102 69 760
11.0 7.4 81.6
1.01 (0.78-1.31) 0.77 (0.57-1.04) Reference
467 401 48 1373
25.1 23.6 4.2 73.7
286 213 35 689
30.8 25.2 6.3 74.0
1.33 (1.11-1.58) 1.10 (0.91-1.33) 1.52 (0.98-2.38) 1.01 (0.85-1.21)
1227 101
92.4 7.6
634 38
94.4 5.7
1.37 (0.94-2.02) Reference
444 1384
24.3 75.7
204 721
22.1 78.0
0.88 (0.73-1.07) Reference
*Crude odds ratios estimate the risk of urinary tract infection among women with the characteristic compared with women in the reference category.
dures on the face sheet of the hospital discharge papers. Because our focus was postpartum urinary tract infection, we excluded 76 cases of pyelonephritis, which most likely reflected antepartum urinary tract infection. Control subjects (n = 1862) were randomly selected in a 2:1 control subject/study subject ratio from the pool of women in the database without a discharge diagnosis of urinary tract infection. To avoid misclassification, we excluded potential control subjects who may have had urinary tract infections that were misclassified as other conditions. These diagnoses included major puerperal infection (ICD-9 code 670); inflammatory diseases of the ovary, fallopian tubes, or peritoneum (ICD-9 code 614); inflammatory diseases of the cervix, vagina, or vulva (ICD-9 code 616); other sexually transmitted diseases (ICD-9 codes 090-099.9); pyelonephritis (ICD-9 code 590); and infections of the genitourinary tract in pregnancy (ICD-9 code 646.6). To validate the diagnosis of urinary tract infection in the database and to assess for information regarding catheterization and peripartum antibiotic use, we reviewed charts from 37 women with urinary tract infections who were delivered of their neonates at the University of Washington Medical Center. We examined the effect of hospital procedures on the prevalence of urinary tract infections, such as fetal monitoring, induction or stimulation of labor or both, tocolysis, cesarean delivery, vacuum extraction, manual extraction, and forceps extraction, as potential risk factors.
Internal and external fetal monitoring were not distinguishable in our database. Also, induction and stimulation of labor were combined for the purpose of this analysis because, although labor induction may be a longer procedure, both procedures frequently involve use of epidural anesthesia and bladder catheterization. We posited that these procedures may increase the risk of urinary tract infection because they might alter urinary tract function, result in an increased number of vaginal examinations with or without instrumentation, thereby potentially contaminating the urethra with vaginal or rectal flora, or cause urethral injury. In addition, we considered epidural anesthesia as a potential risk factor; however, this variable was incompletely reported in our database. Demographic factors of interest included race, age, marital status, educational level, income, location of birth setting, parity, smoking status, and alcohol use. Black, Hispanic, and Native American race were combined into 1 race-ethnicity variable because of the small numbers in each group. In addition, the maternal conditions diabetes, renal disease (including acute and chronic renal disease and nephrolithiasis, but excluding infection), obesity, anemia, cephalopelvic disproportion, abruptio placentae, placenta previa, and preeclampsia or eclampsia (a combined diagnosis) were examined as risk factors for postpartum urinary tract infection. These factors reflect maternal health and immune status, as well as the degree of complication associated with delivery. In addition, the association between postpartum urinary
Schwartz et al 549
Volume 181, Number 3 Am J Obstet Gynecol
Table II. Obstetric procedures and complications among women with and without urinary tract infection in Washington State, 1987-1993 Control subjects (n = 1862)
Case subjects (n = 931)
Condition
No.
%
No.
%
Cesarean delivery Fetal monitoring† Labor induction‡ Outlet forceps Tocolysis Manual extraction Vacuum extraction Perineal laceration Episiotomy
367 963 445 171 33 268 150 493 528
20.7 51.7 23.9 9.7 2.7 14.4 8.1 26.5 28.4
374 481 251 81 51 125 55 169 229
41.3 51.7 27.0 9.1 8.4 13.4 5.9 18.2 24.6
Crude odds ratio and 95% confidence interval* 2.70 (2.27-3.20) 1.00 (0.85-1.17) 1.18 (0.98-1.41) 0.93 (0.70-1.22) 3.30 (2.15-5.06) 0.92 (0.73-1.16) 0.72 (0.52-0.99) 0.62 (0.51-0.75) 0.82 (0.69-0.99)
*Crude odds ratios estimate the risk of urinary tract infection among women with as compared with those without the stated condition. †Includes both internal and external monitoring. ‡Includes both induction and stimulation of labor.
tract infection and delivery of a low-birth-weight (<2500 g) infant was explored. The risk of acquiring a postpartum urinary tract infection associated with hospital procedures, maternal medical conditions, and demographic characteristics was estimated by odds ratios derived by using Mantel-Haenszel procedures.8 Statistical analyses were carried out by using SAS (SAS Institute, Cary, NC). Presence of confounding was assessed by comparing crude and adjusted odds ratios obtained from stratified analyses. A variable was considered to be a confounder if it was associated with both the risk factor and urinary tract infection and if the odds ratio adjusted for the confounder differed from the crude odds ratio by ≥10%. We specifically evaluated whether cesarean delivery modified associations between urinary tract infection and hospital procedures, maternal diagnoses, and demographic characteristics. Results Women who acquired urinary tract infection were similar to those who did not acquire uninary tract infection with respect to age, educational level, smoking, alcohol use, income, location of birth, and parity (Table I). Women of black, Hispanic, or Native American race were slightly more likely than white women to have urinary tract infections (odds ratio, 1.30; 95% confidence interval, 1.03-1.64); however, small numbers of women in each race group precluded examining individual race groups as a predictor for the presence of urinary tract infection. Unmarried women also demonstrated a slightly greater risk than married women (odds ratio, 1.33; 95% confidence interval, 1.11-1.58). The frequency of procedures performed differed between patients and control subjects (Table II). Women who were delivered of their infants by cesarean were nearly 3 times more likely to acquire a postpartum urinary tract infection than women who were not (odds
ratio, 2.70; 95% confidence interval, 2.27-3.20), as were women who received tocolytic agents (odds ratio, 3.30; 95% confidence interval, 2.15-5.06). Cesarean delivery did not significantly confound the relationship between tocolysis and postpartum urinary tract infection. Vacuum extraction, perineal laceration, and episiotomy were more common in control subjects than in women with urinary tract infections and initially appeared to be associated with a decreased risk of urinary tract infection (Table II). This association no longer remained after cesarean delivery was controlled for (data not shown). Women with urinary tract infections and control subjects were similar with respect to prevalence of gestational or preexisting diabetes, obesity, and anemia but differed with respect to preexisting renal disease (odds ratio, 3.77; 95% confidence interval, 1.69-8.41), preeclampsia-eclampsia (odds ratio, 3.62; 95% confidence interval, 2.72-4.84), and pregnancies or deliveries complicated by abruptio placentae (odds ratio, 2.17; 95% confidence interval, 1.06-4.44), placenta previa (odds ratio, 3.67; 95% confidence interval, 1.77-7.59), cephalopelvic disproportion (odds ratio, 1.58; 95% confidence interval, 1.09-2.27), or breech presentation (odds ratio, 1.84; 95% confidence interval, 1.32-2.56; Table III). However, few women in both the urinary tract infection and control groups had renal disease, abruptio placentae, or placenta previa. After we adjusted for cesarean section, both renal disease (odds ratio, 3.89; 95% confidence interval, 1.80-8.41) and preeclampsia-eclampsia (odds ratio, 3.21; 95% confidence interval, 2.36-4.38) remained as significant maternal risk factors for postpartum urinary tract infection (Table IV). When stratified by method of delivery, renal disease was a significant risk factor for postpartum urinary tract infection among women who were delivered vaginally (odds ratio, 5.47; 95% confidence interval, 2.04-14.64) but not among
550 Schwartz et al
September 1999 Am J Obstet Gynecol
Table III. Maternal medical conditions and pregnancy complications among women with and without urinary tract infections in Washington State, 1987-1993 Control subjects (n = 1862) Condition Diabetes† Obesity‡ Anemia Renal disease§ Preeclampsiall Cephalopelvic disproportion Abruptio placentae Placenta previa Breech delivery
Case subjects (n = 931)
No.
%
No.
%
Crude odds ratio and 95% confidence interval*
48 60 32 8 72 69 14 10 78
2.6 14.9 2.7 0.5 4.2 4.0 0.8 0.6 4.5
36 30 19 15 120 53 15 18 69
3.9 14.7 3.2 1.8 13.8 6.1 1.7 2.1 8.0
1.52 (0.98-2.35) 0.99 (0.62-1.59) 1.20 (0.68-2.14) 3.77 (1.69-8.41) 3.62 (2.72-4.84) 1.58 (1.09-2.27) 2.17 (1.06-4.44) 3.67 (1.77-7.59) 1.84 (1.32-2.56)
*Crude odds ratios estimate the risk of urinary tract infection among women with as compared with those without the stated condition. †Includes linked information for gestational and other diabetes. ‡Defined as prepregnancy weight of ≥180 pounds. §Includes linked data for acute or chronic renal disease and nephrolithiasis but excludes infection. llIncludes both eclampsia and preeclampsia.
women who were delivered by cesarean (odds ratio, 1.43; 95% confidence interval, 0.24-8.59). The number of women who had both renal disease and a cesarean delivery, however, was small. Labor induction or stimulation was of borderline significance as a risk factor for postpartum urinary tract infection both before and after we adjusted for cesarean delivery. When stratified by method of delivery, women who were delivered vaginally and underwent labor induction or stimulation had a slightly increased risk of urinary tract infection (odds ratio, 1.35; 95% confidence interval, 1.09-1.68), although, among women who were delivered of their infants by cesarean, induction of labor was not associated with urinary tract infection (odds ratio, 0.99; 95% confidence interval, 0.70-1.41). Abruptio placentae was not a significant risk factor for postpartum urinary tract infection after we adjusted for cesarean section (odds ratio, 1.78; 95% confidence interval, 0.93-3.42). However, among women who underwent vaginal delivery, abruptio placentae was associated with an increased risk of urinary tract infection (odds ratio, 5.02; 95% confidence interval, 1.84-13.64). Next, we further explored the observed associations between occurrence of postpartum urinary tract infection with race-ethnicity and marital status. The association between black, Hispanic, or Native American race and urinary tract infection was no longer statistically significant after adjustment by preeclampsia-eclampsia (adjusted odds ratio, 1.25; 95% confidence interval, 0.981.61). The association of unmarried status with postpartum urinary tract infection, however, was not changed after adjustment by either urban birthplace (adjusted odds ratio, 1.35; 95% confidence interval ,1.101.66) or mother’s age (adjusted odds ratio 1.34; 95% confidence interval, 1.11-1.61).
Low-birth-weight infants were more common among women with urinary tract infections than among control subjects (15.2% vs 5.4%; odds ratio, 3.11; 95% confidence interval, 2.40-4.03). Of note, low birth weight was also more common among women who gave birth in an urban setting (9.5% vs 4.3%; odds ratio, 2.32; 95% confidence interval, 1.03-5.20), among unmarried women (12.4% vs 7.2%; OR = 1.81, 95% confidence interval, 1.38-2.37), and among women of black, Hispanic, or Native American race (15.3% vs 7.5%; odds ratio, 2.24; 95% confidence interval, 1.59-3.14). Thirty-seven women with urinary tract infection who gave birth at the University of Washington Medical Center were randomly selected for chart review. Of these, 32 had urinary tract infection diagnosed by urine culture, and the remainder had a diagnosis by urinalysis alone. Finally, we examined the association between length of hospital stay and maternal diagnosis of urinary tract infection. Because the risk of acquiring a urinary tract infection is known to be strongly associated with bladder catheterization and because we did not have specific data as to whether catheterization had been performed, we examined the association of urinary tract infections with length of hospital stay separately among women presumably at high, as opposed to low, catheterization risk. The group at high risk for catheterization included those women undergoing tocolysis or cesarean delivery or those with preeclampsia-eclampsia; those without these conditions were considered to be at low risk for catheterization. Length of stay was significantly associated with urinary tract infection among women both at low (P value for test of trend < .001) and at high (P value for test of trend < .001) likelihood of having had bladder catheterization performed (Table V).
Schwartz et al 551
Volume 181, Number 3 Am J Obstet Gynecol
Table IV. Maternal medical conditions, procedures, and pregnancy complications in Washington State, 1987-1993: Risk estimates for urinary tract infection adjusted for cesarean delivery Control subjects (n = 2128) Condition Renal disease* Cesarean delivery No cesarean delivery Preeclampsia† Cesarean delivery No cesarean delivery Cephalopelvic disproportion Cesarean delivery No cesarean delivery Breech presentation Cesarean delivery No cesarean delivery Placenta previa Cesarean delivery No cesarean delivery Abruptio placentae Cesarean delivery No cesarean delivery Labor induction‡ Cesarean delivery No cesarean delivery
Case subjects (n = 1064) No.
%
Adjusted odds ratio and 95% confidence interval
No.
%
2/328 6/1359
0.6 0.4
3/346 12/507
0.9 2.4
3.89 (1.80-8.41)
18/331 53/1359
5.4 3.9
65/349 52/514
18.6 10.1
3.21 (2.36-4.38)
68/342 1/1383
19.9 0.1
52/347 1/514
15.0 0.2
0.73 (0.49-1.08)
64/347 14/1383
18.4 1.0
64/350 5/514
18.3 1.0
0.99 (0.69-1.41)
9/342 1/1383
2.6 0.1
16/347 2/514
4.6 0.4
2.00 (0.94-4.28)
8/342 6/1383
2.3 0.4
4/347 11/514
1.2 2.1
1.78 (0.93-3.42)
78/367 361/1405
21.2 25.7
79/374 169/531
21.1 31.8
1.24 (1.03-1.49)
*Renal disease combines data for acute or chronic renal disease or nephrolithiasis but excludes infection. †Includes both eclampsia and preeclampsia. ‡Includes both induction and stimulation of labor.
Comment This is a large case-control study that confirms the data of 2 prior studies9, 10 demonstrating cesarean delivery to be an important risk factor for postpartum urinary tract infection. These data also identify 2 additional procedures, tocolysis and induction of labor, as risk factors for urinary tract infection. Women who undergo these procedures often undergo urethral catheterization, which is estimated to result in urinary tract infection in 10% of patients.11, 12 Maternal renal disease, preeclampsiaeclampsia, and unmarried status were also risk factors for postpartum urinary tract infection. Additionally, length of hospital stay was also associated with risk of urinary tract infection. Identification of cesarean delivery, tocolysis, and induction of labor as urinary tract infection risk factors may underscore the prominence of bladder catheterization as a cause of nosocomial urinary tract infection. Unfortunately, the information in this database did not identify women who had received bladder catheterization. Even if catheterization data had been available, the catheterization risk for postpartum urinary tract infection may still have been difficult to separate from risk associated with tocolysis, cesarean delivery, or induction of labor because many of these women also would have had catheterization. It may be that the risk of urinary tract infection conferred by these procedures is actually mediated or caused by bladder catheterization. Prospective,
controlled studies as to the presence and length of continuous versus intermittent catheterization would be helpful. Prior studies have reported maternal risk factors for postcesarean infectious complications, including urinary tract infection. In a prospective study Parrott et al13 found obesity and low socioeconomic status to be risk factors for postcesarean urinary tract infection. Nielsen and Hokegard14 found obesity, anemia, duration of labor, and duration of ruptured membranes to be associated with an increased risk of postcesarean nosocomial infection. Unlike these investigators, we did not find anemia, obesity, diabetes, or income to be associated with occurrence of urinary tract infection. There may be several explanations for the greater than 3-fold risk of urinary tract infection in women with renal disease or preeclampsia. Both preeclampsia and renal disease are frequently associated with proteinuria, which may enhance a woman’s risk for acquiring a urinary tract infection.15 Renal disease could also signify anatomic abnormalities (eg, reflux nephropathy, scarring, or nephrolithiasis providing a nidus for infection) or loss of immunoglobulins in the urine contributing to a relatively immunosuppressed state. Additionally, women with either renal disease or preeclampsia may require bladder catheterization more frequently than control subjects. Many of the maternal conditions or complications of delivery frequently requiring cesarean delivery, such as
552 Schwartz et al
September 1999 Am J Obstet Gynecol
Table V. Maternal urinary tract infection and length of hospital stay in women in Washington State, 1987-1993, stratified by risk of bladder catheterization Control subjects (n = 1862)* Length of stay (d) Low catheterization risk‡ 1 2 3 4 ≥5 High catheterization risk§ ≤2 3 4 ≥5
Study subjects (n = 931)
No.
%
No.
%
Odds ratio† and 95% confidence interval
466 597 174 31 16
36.4 46.4 13.6 2.4 1.2
101 199 113 48 96
18.2 35.7 20.3 8.6 17.2
Reference 1.54 (1.17-2.03) 3.00 (2.15-4.18) 7.14 (4.21-12.15) 27.68 (15.16-51.19)
20 129 124 47
6.3 40.3 38.8 14.7
6 83 120 165
1.6 22.2 32.1 44.1
Reference 2.14 (0.77-6.25) 3.23 (1.17-9.32) 11.70 (4.13-34.77)
*Two hundred fifty-eight control subjects were missing data on length of hospital stay. †Odds ratios compare risk of urinary tract infection with that in women in the specified reference category. ‡Includes women who did not have preeclampsia and did not undergo tocolysis or cesarean delivery. §Includes women who had preeclampsia or underwent tocolysis or cesarean delivery; 1 and 2 days were combined because no patients at high risk for catheterization stayed only 1 day.
cephalopelvic disproportion, breech delivery, and placenta previa, were associated with urinary tract infection by univariate analysis. However, when adjusted for cesarean delivery, these procedures were not found to be significant risk factors for urinary tract infection. This suggests that the risk thought to be associated with these procedures was presumably caused by catheterization performed as a result of cesarean delivery. Although cesarean delivery potentially confounded the relationship between abruptio placentae and urinary tract infection, abruptio placentae was a risk factor for postpartum urinary tract infection in women who underwent vaginal delivery. Many of these women may have undergone catheterization as a result of being at bed rest or for monitoring maternal volume status and urine output. The association between delivery of a low-birth-weight infant and postpartum urinary tract infection, which has not been previously reported, may be attributed to a complex interplay of demographic factors. However, we did not find that race or marital status confounded the association between low birth weight and the presence of a urinary tract infection. Asymptomatic bacteriuria of pregnancy, a known risk factor for low birth weight and prematurity,6, 16-19 is found more often in women of low socioeconomic status.3, 4 Furthermore, the association between unmarried status and urinary tract infections, also observed by Hooton et al,20 may reflect behavioral factors, such as frequency of sexual intercourse or exposure to new partners. The relationship between length of hospital stay and maternal diagnosis of urinary tract infection has important implications and deserves further study. A retrospective cohort study of women who underwent cesarean delivery found that urinary tract infection significantly prolonged hospital stay (7.2 days compared with 5.7 days, P < .001).21 Although a postpartum urinary tract infec-
tion may result in a longer hospital stay, it may also be that the nosocomial risks associated with longer hospital stays, such as use of an indwelling bladder catheter, increase the incidence of postpartum urinary tract infection. Additionally, a longer hospital stay allows a greater opportunity for a clinician to make a urinary tract infection diagnosis. We attempted to adjust for factors that reflect a complicated delivery, including cesarean, tocolysis, and low birth weight, and found that the association remained. However, this case-control analysis does not allow a cause-and-effect determination between urinary tract infection and length of hospital stay. Other limitations of our study result from the use of birth certificate data and hospital discharge data, which were not specifically designed or originally collected for research purposes. The hospital discharge record, for example, lists only 5 diagnoses and 3 procedures. Parrish et al22 examined the accuracy of birth certificate and hospital discharge data in Washington State and detected a 94% match between the birth certificate and the mother’s and infant’s hospital discharge abstracts. In the same study the ability of this data source to correctly identify obstetric procedures, including cesarean delivery, was found to be >90% for most types of delivery. However, it was also found that obstetric procedures not related to reimbursement, such as induction of labor and episiotomy, were not reliably reported in the database. Next, criteria for the diagnosis of urinary tract infections may have varied among our study subjects. Our chart review at one urban tertiary medical center in Washington State indicates that most women given a diagnosis of urinary tract infection in this population had the diagnosis made on the basis of urinalysis and culture. If less stringent criteria were used to make a urinary tract infection diagnosis, this misclassification would produce risk estimates lower than the true ones. On the other
Volume 181, Number 3 Am J Obstet Gynecol
hand, some women with urinary tract infection, such as those who did not have a urinalysis or culture or those occurring after hospital discharge, may have been missed. By excluding pyelonephritis from our case definition, other potential study subjects may have been missed as well. In addition, some subjects in our study may not have had postpartum urinary tract infection but rather antepartum asymptomatic bacteriuria or symptomatic urinary tract infection. These types of misclassification errors may also have resulted in underestimation of the true odds ratios. Despite the limitations, this study contributes important information. First, our findings demonstrate the risk of urinary tract infection associated with cesarean delivery. Second, this study identifies tocolysis as an additional risk factor. Although we could not directly examine bladder catheterization as a risk factor for nosocomial urinary tract infection, the procedures that we found to be risk factors often require bladder catheterization, thus indirectly highlighting its importance. Third, renal disease and preeclampsia were identified as additional risk factors that may also be associated with catheterization. Fourth, we report a strong association between postpartum urinary tract infection and length of hospital stay. These findings reinforce the practice of avoiding bladder catheterization when possible and further delineate medical conditions that might predispose a woman to postpartum urinary tract infection. Further research to confirm our findings might include information on prepartum and postpartum urinalyses and cultures, as well as data on bladder catheterization, epidural anesthesia, and reason for prolonged hospital stay. Attention to such factors could decrease maternal morbidity and cost in the immediate postpartum period. We thank Dr Pat Starzyk of the Washington State Center for Health Statistics for providing the data, Mr Bill O’Brien for assistance with data management, and Drs Beth Mueller, Steve Schwartz, and Walter Stamm for their advice and support. REFERENCES
1. Ronald AR, Pattullo ALS. The natural history of urinary tract infection in adults. Med Clin North Am 1991;75:299-312. 2. Weissenbacher ER, Reisenberger K. Uncomplicated urinary tract infections in pregnant and non-pregnant women. Curr Opin Obstet Gynecol 1993;5:513-6.
Schwartz et al 553
3. Turck M, Goffe BS, Petersdorf RG. Bacteriuria of pregnancy. N Engl J Med 1962;266:857-60. 4. Williams GL, Campbell H, Davies KJ. The influence of age, parity, and social class on the incidence of asymptomatic bacteriuria in pregnancy. J Obstet Gynaecol Br Commonw 1969;76:22939. 5. Layton R. Infection of the urinary tract in pregnancy: an investigation of a new routine in antenatal care. J Obstet Gynaecol Br Commonw 1964;71:927-33. 6. Savage WE, Hajj SN, Kass EH. Demographic and prognostic characteristics of bacteriuria in pregnancy. Medicine 1967;46:385-407. 7. Chng PK, Hall MH. Antenatal prediction of urinary tract infection in pregnancy. Br J Obstet Gynaecol 1982;89:8-11. 8. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22:719-48. 9. Buchholz N-P, Daly-Grandeau E, Huber-Buchholz M-M. Urological complications associated with caesarean section. Eur J Obstet Gynecol Biol 1994;56:161-3. 10. Leigh DA, Emmanuel FXS, Sedgwick J, Dean R. Post-operative urinary tract infection and wound infection in women undergoing caesarean section: a comparison of two study periods in 1985 and 1987. J Hosp Infect 1990;15:107-16. 11. Johnson JR, Roberts PL, Olsen RJ, Moyer KA, Stamm WE. Prevention of catheter-associated urinary tract infection with a silver oxide-coated urinary catheter: clinical and microbiologic correlates. J Infect Dis 1990;162:1145-50. 12. Stamm WE. Catheter-associated urinary tract infections: epidemiology, pathogenesis, and prevention. Am J Med 1991;91(Suppl 3B):65S-71. 13. Parrott T, Evans AJ, Lowes A, Dennis KJ. Infection following caesarean section. J Hosp Infect 1989;13:349-54. 14. Nielsen TF, Hokegard K-H. Postoperative cesarean section morbidity: a prospective study. Am J Obstet Gynecol 1983;146:911-6. 15. Cruikshank DP, Hays PM. Maternal physiology in pregnancy. In: Gabbe SG, Niebyl JR, Simpson JL, editors. Obstetrics: normal and problem pregnancy. New York: Churchill Livingstone; 1991. p. 125-46. 16. Kincaid-Smith P, Bullen M. Bacteriuria in pregnancy. Lancet 1965;1:395-9. 17. DeBaun M, Rowley D, Province M, Stockbauer JW, Cole FS. Selected antepartum medical complications and very-low-birthweight infants among black and white women. Am J Public Health 1994;84:1495-7. 18. Schieve LA, Handler A, Hershow R, Persky V, Davis F. Urinary tract infection during pregnancy: its association with maternal morbidity and perinatal outcome. Am J Public Health 1994;84:405-10. 19. Kass EH. Pyelonephritis and bacteriuria: a major problem in preventive medicine. Ann Intern Med 1962;56:46-53. 20. Hooton TM, Scholes D, Hughes JP, et al. A prospective study of risk factors for symptomatic urinary tract infection in young women. N Engl J Med 1996;335:468-74. 21. Henderson E, Love EJ. Incidence of hospital-acquired infections associated with caesarean section. J Hosp Infect 1995;29:245-55. 22. Parrish KM, Holt VL, Connell FA, Williams B, LoGerfo JP. Variations in the accuracy of obstetric procedures and diagnoses on birth records in Washington State, 1989. Am J Epidemiol 1993;138:119-27.