Professional Documents
Culture Documents
Risk Factors For Postpartum Urinary Retention: A Systematic Review and Meta-Analysis
Risk Factors For Postpartum Urinary Retention: A Systematic Review and Meta-Analysis
Risk Factors For Postpartum Urinary Retention: A Systematic Review and Meta-Analysis
x
Systematic review
www.bjog.org
Background Postpartum urinary retention (PUR) is a common risk factors of overt and covert PUR and calculated pooled odds
condition with varying prevalence. Measurement of the post-void ratios (ORs) with 95% confidence intervals.
residual volume (PVRV) is not regularly performed. Various
Main results Twenty-three observational studies with original data
studies have been published on overt (the inability to void after
were eligible for data extraction, of which 13 could be used for
giving birth, requiring catheterisation) and covert (an increased
meta-analysis. Statistically significant risk factors for overt PUR
PVRV after spontaneous micturition) PUR. To evaluate which
were epidural analgesia (OR 7.7), instrumental delivery (OR 4.5),
clinical prognostic factors are related to PUR, the identification of
episiotomy (OR 4.8) and primiparity (OR 2.4). For covert PUR,
independent risk factors for covert and overt PUR is needed.
variety in the definitions used resulted in heterogeneity; no
Objectives We performed a systematic review and meta-analysis significant prognostic factors were found.
of observational studies reporting on risk factors for PUR.
Conclusions Instrumental delivery, epidural analgesia, episiotomy
Search strategy Systematic search of MEDLINE and EMBASE to and nulliparity are statistically significantly associated with a
September 2011. higher incidence of overt PUR. The same factors were identified
for covert PUR, but without statistical significance. Uniformity in
Selection criteria Articles that reported on women diagnosed with
definitions in future research is essential to create a prognostic
PUR or with an abnormal PVRV.
model.
Data collection and analysis The included articles were selected
Keywords Meta-analysis, postpartum period, risk factors,
by two authors. We constructed two-by-two tables for potential
systematic review, urinary retention, voiding dysfunction.
Please cite this paper as: Mulder F, Schoffelmeer M, Hakvoort R, Limpens J, Mol B, van der Post J, Roovers J. Risk factors for postpartum urinary retention:
a systematic review and meta-analysis. BJOG 2012;119:1440–1446.
1440 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG
Risk factors for postpartum urinary retention
reported prevalence for covert PUR varies widely In addition, papers with urinary retention as a secondary
(1.5–45%),7,8 the consequences of this condition are still outcome were used for analysis.
debated.1,7 As screening for post-void residual volumes Two authors (FEMM and MAS) independently assessed
(PVRVs) after delivery is seldom part of standard postpar- the eligibility of the studies and extracted available data. To
tum care, the management of covert PUR only ‘exists’ in assess the quality of all included studies, the STROBE
study designs. In order to evaluate the clinical need for the guidelines were used.15 Any disagreements between the two
treatment of covert PUR, clinical factors related to PUR reviewers were resolved through discussion.
must be identified. Prognostic factors that are associated For the included studies, a two-by-two table was created
with PUR are the duration of labour, instrumental-assisted to classify potential risk factors for PUR. When we were
delivery, episiotomies, birthweight, parity and epidural unable to construct a two-by-two table because of missing
anaesthesia.4,5,8–13 Although these potential risk factors seem data, we contacted the authors of the original paper for
to be highly related, it is necessary to evaluate independent additional data. We calculated odds ratios (ORs) and 95%
prognostic factors, and thus whether women with multiple confidence intervals (CIs) from each two-by-two table and
clinical risk factors have an increased risk of PUR. constructed forest plots using Review Manager version
The aim of this review was to identify and quantify 5.0.2 (Review Manager (RevMan) [Computer program].
clinical factors that can predict the occurrence of PUR, and Version 5.0.2. Copenhagen: The Nordic Cochrane Centre,
to quantify their influence as independent risk factors. The Cochrane Collaboration, 2009). To quantify heteroge-
neity, i.e. to determine the proportion of total variance
between different studies explained by heterogeneity, the I2
Methods
with a chi-squared test was used to calculate the P values.
We performed a systematic search to identify peer-reviewed A fixed-effect model was used when heterogeneity was low
publications on PUR. We searched the electronic databases to moderate (I2 < 50%). When homogeneity was rejected
MEDLINE (OVID) from 1948 and EMBASE (OVID) from and thus heterogeneity was substantial to high, we used a
1980 to September 2011 without applying limits (including random effect model (I2 ‡ 50%).16
language and publication type). We excluded animal studies
by using double negation (i.e. in MEDLINE: not animals/
Results
not humans/). The search strategy consisted of free-text
words (tw) and Subject Headings (MeSH, SH: command/) The OVID Medline search (1948 to September 2011)
related to: (1) urinary retention (bladder retention, bladder retrieved 306 records, 289 of which were unique, whereas
and voiding functions or dysfunctions); and (2) delivery, the EMBASE search (1980 to September 2011) retrieved
the postpartum period and obstetric procedures that 605 records. In total, 696 unique records were obtained
enhance the chance of urinary retention (i.e. episiotomy, from MEDLINE and EMBASE (Figure 1).14 For data
caesarean, anaesthesia). Synonyms for 1 and 2 were com- extraction, 23 original papers met the inclusion criteria and
bined with the boolean operator ‘or’, whereas search sets 1 were selected in this systematic review for analysis.
and 2 were combined with the boolean operator ‘and’ (see Studies in which a two-by-two table could not be con-
Appendix S1 for the MEDLINE search strategy).14 The elec- structed were excluded. Missing data were verified by con-
tronic search was supplemented by a hand search of refer- tacting the authors of the original articles. In total, eight
ence lists of relevant articles. We searched for relevant authors were contacted. Four authors sent additional data,
(unpublished) studies in the grey literature in the following allowing the construction of a two-by-two table. Three
databases: Inside Conferences, Systems for Information in authors responded that their data were no longer available
Grey Literature and Clinical Trials.gov. and one author did not respond despite several reminders.
The search included an iterative process to refine the Only studies with complete datasets per risk factor were
search strategy through the addition of search terms as new included in the analysis.
relevant citations were identified. The bibliographic records A distinction was made between overt and covert PUR.
retrieved were downloaded and imported into Reference The data of 13 studies were finally used for the meta-
Manager software (Reference Manager, version 12.0, analysis.2–5,7,8,10–13,17–19 Of these studies, nine were pro-
Thomson Reuters, 2008, New York, USA) to de-duplicate, spective2,3,5,7,8,11–13,19 and four were retrospective.4,10,17,18
store and analyse the search results. Parity was reported in five studies for overt PUR, 3,4,10,17,18
Inclusion criteria were studies that reported women diag- and in three studies for covert PUR.5,8,11 Instrumental deliv-
nosed with PUR or women with an abnormal PVRV. ery was reported in five studies for overt PUR,2–4,17,18 and in
Women with pre-existing kidney disease or urinary tract five studies for covert PUR.5,7,8,12,13 Epidural analgesia was
problems were excluded. Only papers presenting original reported in seven studies for overt PUR,2–4,10,17–19 and in
work regarding urinary retention postpartum were included. two studies for covert PUR.8,12 Episiotomy and/or vaginal
ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 1441
Mulder et al.
Identification
MEDLINE EMBASE
(n = 289) (n = 605)
Records screened
Records excluded (n = 110)
(n = 198)
Prospective studies (n = 9)
Studies included in Cohort (n = 7)
quantitative synthesis Cross sectional (n = 1)
(meta-analysis) RCT (n = 1)
(n = 13) Retrospective studies (n = 4)
Cohort (n = 2)
Case control (n = 2)
1442 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG
Risk factors for postpartum urinary retention
Figure 2. Forest plots for risk factors for overt postpartum urinary retention (PUR).
PUR was 1.1 (95% CI 0.5–2.6). Two studies investigating between episiotomy and covert PUR was mixed (pooled
the effect of epidural analgesia showed ORs of 1.2 and OR 1.5; 95% CI 0.8–2.2), as was the comparison
4.7, respectively, in favour of no epidural analgesia between parity and covert PUR (OR 1.3; 95% CI
(pooled OR 2.3; 95% CI 0.7–7.7). The association 0.8–2.2).
ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 1443
Mulder et al.
A
Instrumental delivery Non-instrumental delivery Odds ratio Odds ratio
Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI M-H, Random, 95% CI
Andolf 2 32 6 507 14.7% 5.57 [1.08, 28.76]
Demaria 7 33 48 121 23.1% 0.41 [0.16, 1.02]
Hee 7 14 16 37 19.1% 1.31 [0.38, 4.50]
Ismail 2 11 35 89 15.2% 0.34 [0.07, 1.68]
Kekre 24 152 60 619 27.9% 1.75 [1.05, 2.91]
Figure 3. Forest plots for risk factors for covert postpartum urinary retention (PUR).
As a result of the absence of studies with multivariate systematically collecting available data and translating the
analyses, we were unable to explore which clinical factors results into ORs, we have created a supplement on the
have been identified as independent risk factors. Conse- available literature and daily clinical practice.
quently, we could not quantify the effect of the various risk With regard to overt PUR, instrumental delivery was the
factors and evaluate whether multiple prognostic factors in only clinical factor with an excellent homogeneity
one woman result in an increased risk of diagnosis with (I2 = 0%). The other identified risk factors (parity, epidural
PUR. analgesia and episiotomy) were obtained from studies that
were highly heterogeneous (I2 = 69–87%). We believe that
this heterogeneity is a result of three factors, namely the
Discussion
use of different definitions, the use of different cut-off val-
Our review identified 23 published studies and a variety ues of PVRV and the variety of study designs. In order to
of clinical risk factors which can lead to PUR.8,20,21 By explore the influence of study design on the results, we also
1444 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG
Risk factors for postpartum urinary retention
performed meta-analyses without the retrospective stud- to a variety of registered patient information. Conse-
ies.4,10,17 Figure 2 illustrates that only the studies from Car- quently, a uniform definition of PUR when conducting
ley et al.,4 Fedorkow et al.10 and Musselwhite et al.17 future studies is mandatory. These data would enable the
showed statistically non-significant ORs. When removing creation of a risk profile to identify women with increased
these retrospective studies from the meta-analyses, the risk of PUR, and open up the possibility for intervention
homogeneity increased considerably (I2 = 0–33%); how- trials for covert PUR.
ever, the pooled ORs were not altered by this exclusion. Our study has some weaknesses that must be acknowl-
Because, in the current meta-analyses, all point estimates edged. First, we were unable to complete our collected data,
show a positive correlation, we feel that the inclusion of although we contacted authors and even co-authors when
the retrospective studies is legitimate. data were missing. However, for 67% of the included stud-
With regard to covert PUR, the included clinical prog- ies, the dataset was complete, and for the remaining studies
nostic factors were obtained from heterogeneous studies we were still able to collect the majority of data. Therefore,
(Table 1). Although they were all prospective studies, point we feel that the missing data did not skew our results too
estimates for instrumental delivery and episiotomy appear greatly. In addition to our analysed risk factors, birthweight
on both sides of the no-effect line (Figure 3A, C). In addi- was also mentioned in two studies to be of significant influ-
tion, epidural analgesia and primiparity had no statistically ence on the prevalence of PUR.7,23 As the authors were
significant effect on covert PUR (Figure 3B, D). unable to provide original data on birthweight, we could
The diversity of our data makes it difficult to translate not perform a meta-analysis for this factor.
our results into guidelines for daily clinical practice. Never- Second, in our meta-analysis, we decided to analyse all
theless, we still decided to present our data with pooled available data for covert PUR. Despite the high heterogene-
ORs because our meta-analyses illustrate this variety ity of the data (mainly caused by PVRV being measured
between studies and, with that, the need for uniformity in at different time points after delivery), we feel that the
definitions for covert PUR. pooling of the available data provides valuable information
The relationship between the different identified risk fac- about the different clinical prognostic factors for covert
tors for covert and overt PUR is still indistinct. The ques- PUR. Therefore, we used a random effect model to calcu-
tion is whether instrumental delivery is a true risk factor or late the pooled ORs. The paucity of current knowledge on
whether this factor is confounded by other factors, such as this topic increases the clinical value of our review.
prolonged labour, epidural analgesia, parity and episiot- Finally, because of the inability to perform multivariate
omy. In addition, the reasons to terminate labour could be analysis, we have been unable to identify independent risk
a confounder. Future studies assessing all factors related to factors and to quantify their individual influence. There-
instrumental delivery are needed to answer the question of fore, we cannot define a specific group of women who are
whether or not instrumental delivery is a true risk factor. at risk for PUR and should, for example, be routinely
Multivariate analysis would be a valuable addition to dis- screened after giving birth.
tinguish between confounders and independent prognostic In conclusion our meta-analyses show that instrumental
factors. In 1997, Yip et al.1 conducted a study to assess the delivery, epidural analgesia, parity and episiotomy can be
influence of obstetric factors on PUR. They performed a identified as clinical risk factors for overt or covert PUR.
multivariate analysis and found that only the duration of As homogeneity is lacking, especially in studies concerning
labour had a statistically significant influence on the preva- covert PUR, uniformity of definition is necessary for future
lence of PUR. Later, they stated that a duration of labour studies with multivariate analysis.
longer than 700 minutes was predictive for PUR.22 Because,
in other studies, data on the duration of labour were lack- Disclosure of interests
ing, we have not been able to create a model for indepen- None.
dent prognostic clinical factors for PUR, and the results of
Yip et al. remain unaltered. Contribution to authorship
A method for the identification of independent risk fac- FEMM, RAH, BWJM, JAMvdP and JPWRR conceived the
tors and the creation of a model for PUR would be to per- idea for the study. JL developed and performed the litera-
form individual patient data (IPD) meta-analysis. By ture search. FEMM and MAS collected the data and
analysing the original databases from each researcher and assessed the quality. FEMM, JPWRR and BWJM performed
combining their data, one could create a reliable prognostic statistical analyses. FEMM, JAMvdP and JPWRR drafted
model. However, one would probably encounter the same the article. All authors reviewed and edited the article.
problems for covert PUR as we did in our meta-analyses,
because different authors have adopted different cut-off Details of ethics approval
values, measurement methods and time points, in addition Not required.
ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG 1445
Mulder et al.
1446 ª 2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology ª 2012 RCOG