Beta-Blockers increase the risk of being born small for gestational age or of being institutionalised during infancy

Authors

  • R-h Xie,

    1. Department of Obstetrics and Gynaecology, Nanfang Hospital, Southern Medical University, Guangzhou, China
    2. OMNI Research Group, Department of Obstetrics and Gynaecology, University of Ottawa Faculty of Medicine, Ottawa, ON, Canada
    3. Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada
    Search for more papers by this author
  • Y Guo,

    1. OMNI Research Group, Department of Obstetrics and Gynaecology, University of Ottawa Faculty of Medicine, Ottawa, ON, Canada
    2. Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada
    Search for more papers by this author
  • D Krewski,

    1. McLaughlin Centre for Population Health Risk Assessment, Institute of Population Health, University of Ottawa, Ottawa, ON, Canada
    2. Risk Sciences International, Ottawa, ON, Canada
    3. Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, ON, Canada
    Search for more papers by this author
  • D Mattison,

    1. McLaughlin Centre for Population Health Risk Assessment, Institute of Population Health, University of Ottawa, Ottawa, ON, Canada
    2. Risk Sciences International, Ottawa, ON, Canada
    Search for more papers by this author
  • MC Walker,

    1. OMNI Research Group, Department of Obstetrics and Gynaecology, University of Ottawa Faculty of Medicine, Ottawa, ON, Canada
    2. Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada
    3. Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, ON, Canada
    Search for more papers by this author
  • K Nerenberg,

    1. Department of Medicine, Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada
    Search for more papers by this author
  • SW Wen

    Corresponding author
    1. OMNI Research Group, Department of Obstetrics and Gynaecology, University of Ottawa Faculty of Medicine, Ottawa, ON, Canada
    2. Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada
    3. Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, ON, Canada
    4. School of Public Health, Central South University, Changsha, China
    • Correspondence: Dr SW Wen, OMNI Research Group, Department of Obstetrics and Gynecology, University of Ottawa, 501 Smyth Road, Box 241, Ottawa, ON K1H 8L6, Canada. Email swwen@ohri.ca

    Search for more papers by this author

Abstract

Objective

To compare infant outcomes between mothers with hypertension treated by beta-blockers alone and by methyldopa alone during pregnancy.

Design

Historical cohort study.

Setting

Saskatchewan, Canada.

Population

Women who delivered a singleton birth in Saskatchewan during the periods from 1 January 1980 to 30 June 1987 or from 1 January 1990 to 31 December 2005 (women who delivered between 1 July 1987 and 31 December 1989 were excluded because the information recorded on maternal drug use during pregnancy is incomplete) with a diagnosis of a hypertensive disorder during pregnancy, and who were dispensed only beta-blockers (n = 416) or only methyldopa (n = 1000).

Methods

Occurrences of adverse infant outcomes were compared between women who received beta-blockers only and women who received methyldopa only during pregnancy, first in all eligible women, and then in women with chronic hypertension and in women with gestational hypertension or pre-eclampsia/eclampsia, separately. Multiple logistic regression analyses were performed to adjust for potential confounding.

Main outcome measures

Small for gestational age (SGA) < 10th percentile, SGA < 3rd percentile, preterm birth, stillbirth, institutionalisation for respiratory distress syndrome (RDS), sepsis, seizure during infancy, and infant death.

Results

Adjusted odds ratios (aORs) and 95% confidence intervals (95% CIs) for infants born to mothers with chronic hypertension who were dispensed beta-blockers only, as compared with infants born to mothers who were dispensed methyldopa only, during pregnancy were: 1.95 (1.21–3.15), 2.17 (1.06–4.44), and 2.17 (1.09–4.34), respectively, for SGA < 10th percentile, SGA < 3rd percentile, and being institutionalised during infancy.

Conclusions

For infants born to mothers with chronic hypertension, compared with those treated by methyldopa alone, those treated by beta-blockers appear to be at increased rates of SGA and hospitalisation during infancy.

Introduction

Beta-blocking and alpha-blocking drugs are commonly used to treat hypertension during pregnancy.[1-4] Frequently used beta-blockers include acebutolol, atenolol, labetalol, mepindolol, metoprolol, pindolol, oxprenolol, and propranolol; clonidine and (mostly) methyldopa are used as alpha agonists.[1-4] Previous studies have suggested that pregnancy exposure to beta-blockers may be associated with low birthweight.[5, 6] In a systematic review of 13 clinical trials, Magee and Duley compared birth outcomes between pregnancies exposed to oral beta-blockers with those exposed to methyldopa, and found that the overall risk of small for gestational age (SGA) in pregnancies using beta-blockers was significantly higher than in those using methyldopa, although the increased risk became nonsignificant after the exclusion of a small trial involving women with exposure to atenolol in early pregnancy.[6] The objective of the present study was to further examine the effect of pregnancy exposure to beta-blockers compared with methyldopa, the other most frequently recommended antihypertensive drug prescribed during pregnancy that acts via a different mechanism.

Methods

We received ethical approval to conduct this study from the Ottawa Hospital Research Ethical Board (2008471-01H, approved on 20 October 2011). The study is based on data from the linked maternal/infant databases in the Canadian Province of Saskatchewan. The study population comprised all women covered by the Saskatchewan prescription drug plan (outpatient prescriptions) and who delivered a singleton birth in Saskatchewan during the periods from 1 January 1980 to 30 June 1987, or from 1 January 1990 to 31 December 2005 (women who delivered between 1 July 1987 and 31 December 1989 were excluded because the information recorded on maternal drug use during pregnancy is incomplete), with a diagnosis of a hypertensive disorder in pregnancy (identified through the International Classification of Diseases ICD–9/ICD–10–CA codes recorded in the database) who were dispensed beta-blockers only or methyldopa only. Women with comorbidity (diabetes, renal disease, and cardio disease) or who were prescribed atenolol were excluded. As professional organisations/societies have recommended against the use of atenolol during pregnancy,[2, 3] because of concerns of fetal harm,[7] an exclusion of this drug from the study will render the results more relevant to current conditions of practice.

The hypertensive disorders in pregnancy examined in this study included: pre-existing hypertension (either chronic hypertension that predated pregnancy or hypertension diagnosed before 20 weeks of gestation); gestational hypertension (hypertension that developed after 20 weeks of gestation and urinary protein excretion of <0.3 g/day); pre-eclampsia (hypertension diagnosed after 20 weeks of gestation and urinary protein excretion of more than 0.3 g/day); and pre-existing hypertension with superimposed pre-eclampsia.[8]

The following data were abstracted from the Saskatchewan health databases: a unique patient identifier, the number of days between the drug dispensing date and the infant birth date, gestational age category (0–23, 24–29, 30–37, 38–41, and more than 42 weeks of gestation), the specific name of beta-blocker or methyldopa, demographic characteristics of the mother, and perinatal outcomes in the infant. Information on the use of prescription drugs to treat antihypertensive disorders during pregnancy was ascertained for each study participant, using the number of days between the date of dispensing the drug and the date of birth of the infant.

Outcomes of interest included: SGA, defined as a birthweight of less than the tenth percentile of the Canadian reference, recommended by Kramer et al.;[9] severe SGA (a birthweight of less than the third percentile); preterm birth (live infants delivered at <37 completed weeks of gestation); stillbirth (stillbirth with a birthweight >500 g or at a gestational age of more than 20 weeks); hospitalisation for respiratory distress syndrome (RDS), sepsis, or seizures during the first year of life (any occurrence); and infant death (death of a live-born infant within the first year of life).

We first compared maternal characteristics and the occurrence of adverse perinatal outcomes between the two study groups: women who received beta-blockers only and women who received methyldopa only during pregnancy. Multiple logistic regression analyses were then performed to assess the independent association between the type of antihypertensive drug and adverse perinatal outcomes, with methyldopa as the reference. Potential confounding variables included in the regression models were: maternal age (<25, 25–29, and ≥30 years, with <25 years as the reference); year of childbirth (1980–1984, January 1985–June 1987, 1990–1994, 1995–1999, 2000–2005, with 1980–1984 as the reference); Saskatchewan assistance plan coverage (no versus yes, with no as the reference); parity (1 versus ≥2, with 1 as the reference); and type of hypertensive disorders in pregnancy (chronic hypertension, including chronic hypertension with superimposed pre-eclampsia, gestational hypertension, and pre-eclampsia/eclampsia, with gestational hypertension as the reference). Confounding variables and their categorisation included in the regression model were determined a priori by an expert panel consisting of two obstetricians, an internist with focus on hypertensive disorders in pregnancy, a population risk scientist, a nurse scientist, a statistician, and a perinatal epidemiologist, based on literature review. Because of the limited sample size, important confounding factors may not be significant between the two study groups. Some important variables were not available in the database. For other variables, although a finer categorisation (e.g. for extremes in age, such as those <18 years of age or those >35 years of age) may be preferable, the numbers in these categories were extremely small, thereby preventing a meaningful analysis. There was an interruption in data availability between 1 July 1987 and 31 December 1989. As a result, data quality may be affected for the years before and after that period. We have therefore categorised ‘year of birth’ into different periods, including the years before and after the interruption. We are also aware that ICD–9 was changed to ICD–10 at the end of 1990s or early in the 2000s. To adjust for the potential impact of changed ICD coding, we have used two further categories: 1995–1999 and 2000–2005 for the variable of ‘year of birth’. A fully adjusted model comprising all independent variables was used in the analysis.

As there are major differences in the clinical indications and duration of treatment of hypertensive disorders of pregnancy, we performed the following two a priori analyses. First, we compared perinatal outcomes between the two study groups in women with a diagnosis of chronic hypertension and in women with a diagnosis of gestational hypertension or pre-eclampsia/eclampsia, separately. This was carried out because gestational hypertension/pre-eclampsia generally develops after 20 weeks of gestation, and antihypertensive drug treatment for these disorders, if any, occurs only late in the second trimester or into the third trimester. On the other hand, antihypertensive treatment of women with chronic hypertension may occur before pregnancy, and may continue from early gestation to late gestation. These differences in treatment regimens may have a direct impact on perinatal outcomes. Second, women with chronic hypertension may be treated with antihypertensive medications for longer durations during pregnancy, and the time of initiation of treatment may be associated with clinical outcomes. As such, we performed an additional multiple logistic regression analysis for women with a diagnosis of chronic hypertension, to adjust for the time to begin the antihypertensive treatment (pre-conception, first trimester, second trimester, and third trimester) as an indicator of duration of therapy. Supplemental analyses restricted to first pregnancies only, and altering the categorisation of maternal age and year of birth, were also performed.

Results

A total of 1416 women were included in the final analysis. Of these, 416 received beta-blockers only and 1000 received methyldopa only during pregnancy.

Table 1 describes the maternal characteristics of the two study groups. Compared with women exposed to methyldopa, women who were dispensed beta-blockers were more likely to be diagnosed with pre-eclampsia and to initiate treatment in later gestation, but were less likely to depend on government assistance.

Table 1. Maternal characteristics of women who received beta-blockers only and methyldopa only during pregnancy, 1980–2005
CharacteristicBeta-blockers only No. (%)Methyldopa only No. (%)Total
Age group
<25 years87 (20.91)234 (23.40)321
25–29 years147 (35.34)312 (31.20)459
≥30 years182 (43.75)454 (45.40)636
Year of birth
1980–198427 (6.49)42 (4.20)69
1985–June 198722 (5.29)35 (3.50)57
1990–199447 (11.30)160 (16.00)207
1995–199966 (15.87)311 (31.10)377
2000–2005254 (61.06)452 (45.20)706
Saskatchewan assistance plan coverage
No391 (94.22)918 (91.80)1309
Yes24 (5.78)82 (8.20)106
Parity
1219 (52.64)523 (52.30)742
≥2197 (47.36)477 (47.70)674
Type of hypertensive disorder
Chronic hypertension179 (43.03)449 (44.90)628
Gestational hypertension203 (48.80)508 (50.80)711
Pre-eclampsia34 (8.17)43 (4.30)77
Initiation of treatment
Preconception102 (24.52)74 (7.40)176
First trimester25 (6.01)80 (8.01)105
Second trimester53 (12.74)180 (18.00)233
Third trimester236 (56.73)666 (66.60)902

Table 2 compares rates of adverse perinatal outcomes between the two study groups. No difference in outcomes between the two study groups was observed.

Table 2. Risk of adverse perinatal outcomes in women receiving beta-blockers relative to those receiving methyldopa during pregnancy, 1980–2005
OutcomeBeta-blockers only No. (%)Methyldopa only No. (%)aOR (95% CI)a
  1. a

    Adjusted for maternal age, parity, birth year, Saskatchewan assistance plan coverage, and type of hypertensive disorder.

  2. b

    SGA: small for gestational age at birth.

  3. c

    Hospitalisation for respiratory distress syndrome, sepsis, or seizure.

SGA (<10th percentile)b72 (17.31)145 (14.50)1.24 (0.90–1.70)
SGA (<3rd percentile)b23 (5.53)55 (5.50)0.99 (0.59–1.66)
Preterm birth101 (24.28)269 (26.90)0.84 (0.64–1.10)
Stillbirth4 (0.96)4 (0.40)2.25 (0.54–9.36)
Hospitalisationc23 (5.53)39 (3.90)1.40 (0.82–2.39)
Infant death2 (0.48)6 (0.60)0.88 (0.17–4.50)

Table 3 compares outcomes between the two study groups in women with a diagnosis of chronic hypertension and in women with a diagnosis of gestational hypertension or pre-eclampsia/eclampsia, separately. For women with chronic hypertension, the risks of SGA < 10th percentile and hospitalisation for RDS, sepsis, and seizures during infancy were higher in infants born to mothers who were dispensed beta-blockers, compared with infants born to mothers who were dispensed methyldopa. For women with gestational hypertension or pre-eclampsia/eclampsia, no statistically significant difference in the incidence of adverse perinatal outcomes was observed between the two study groups, except for preterm birth, for which the rate was higher in the methyldopa group.

Table 3. Risk of adverse perinatal outcomes in women receiving beta-blockers relative to those receiving methyldopa during pregnancy by type of hypertensive disorder in pregnancy, 1980–2005
OutcomeChronic hypertensionGestational hypertension (including pre-eclampsia)
Beta-blockers only No. (%)Methyldopa only No. (%)aOR (95% CI)aBeta-blockers only No. (%)Methyldopa only No. (%)aOR (95% CI)a
  1. a

    Adjusted for maternal age, parity, birth year, and Saskatchewan assistance plan coverage.

  2. b

    SGA: small for gestational age at birth.

  3. c

    Hospitalisation for respiratory distress syndrome, sepsis, or seizure.

SGA (<10th percentile)b38 (21.23)58 (12.92)1.77 (1.11–2.82)34 (14.35)87 (15.79)0.92 (0.59–1.44)
SGA (<3rd percentile)b16 (8.94)23 (5.12)1.67 (0.85–3.30)7 (2.95)32 (5.81)0.52 (0.22–1.24)
Preterm birth50 (27.93)113 (25.17)1.14 (0.77–1.69)51 (21.52)156 (28.31)0.65 (0.45–0.95)
Stillbirth3 (1.68)2 (0.45)3.34 (0.54–20.66)1 (0.42)2 (0.36)0.41 (0.09–21.81)
Hospitalisationc16 (8.94)21 (4.68)2.10 (1.06–4.15)7 (2.95)18 (3.27)0.82 (0.33–2.02)
Infant death2 (1.12)2 (0.45)2.07 (0.26–16.31)0 (0.00)4 (0.73)

Table 4 presents results for women with chronic hypertension, after additional adjustment for the time of initiation of antihypertensive drugs. The rates of SGAs (both <10th percentile and <3rd percentile) and hospitalisation for RDS, sepsis, and seizures during infancy were significantly higher among infants born to mothers who were dispensed beta-blockers, compared with infants born to mothers who were dispensed methyldopa.

Table 4. Risk of adverse perinatal outcomes in women receiving beta-blockers relative to those receiving methyldopa for chronic hypertension during pregnancy 1980–2005
OutcomeChronic hypertension 
Beta-blockers only No. (%)Methyldopa only No. (%)aOR (95% CI)a
  1. a

    Adjusted for maternal age, parity, birth year, Saskatchewan assistance plan coverage, and time of starting treatment (pre-conception, first trimester, second trimester, and third trimester).

  2. b

    SGA: small for gestational age at birth.

  3. c

    Hospitalisation for respiratory distress syndrome, sepsis, or seizure.

SGA (<10th percentile)b38 (21.23)58 (12.92)1.95 (1.21–3.15)
SGA (<3rd percentile)b16 (8.94)23 (5.12)2.17 (1.06–4.44)
Preterm birth50 (27.93)113 (25.17)1.18 (0.79–1.77)
Stillbirth3 (1.68)2 (0.45)2.69 (0.40–18.17)
Hospitalisationc16 (8.94)21 (4.68)2.17 (1.09–4.34)
Infant death2 (1.12)2 (0.45)0.62 (0.05–7.58)

Results from supplemental analyses restricted to first pregnancies or altering the categorisation of maternal age and year of birth were similar, although statistically less stable because of the reduced number of births after the application of the restriction and the finer categorisation of confounding variables (data available upon request).

Discussion

Main findings

This observational study, based on a large population-based retrospective cohort, found that beta-blockers appear to have a strong effect on adverse perinatal outcomes, with an adjusted odds ratio of 1.95 (95% CI 1.21–3.14) for SGA < 10th percentile, 2.17 (95% CI 1.06–4.44) for SGA < 3rd percentile, and 2.17 (95% CI 1.09–4.34) for hospitalisation for RDS, sepsis, and seizures during infancy, as compared with methyldopa, in pregnant women with chronic hypertension.

Strengths and limitations

Our study has several strengths. First, based on a strong biological rationale, the study was designed to test the hypothesis of potential adverse fetal effects of beta-blockers, as animal and in vitro human experiments have shown that beta-blockers may adversely affect fetal circulation.[10-12]. Both selective beta-blockers (metoprolol) and nonselective beta-blockers (propranolol) might adversely affect fetal circulation.[10, 11] Second, we applied a number of approaches to control for potential confounding to ensure that the groups under study were comparable. The analysis was restricted to women with the same type of hypertensive disorder (chronic versus gestational) and who were dispensed with beta-blockers only, versus those who were dispensed with methyldopa only. Extensive adjustment for known clinical predictors of adverse perinatal outcomes by multiple regression analysis reduced the possibility of residual confounding. Study subjects with co-morbidity (including diabetes, renal disease, and cardiac disease) or with pregnancy exposure to atenolol, a beta-blocker with known harmful effects on the fetus, were excluded, to further mitigate possible bias and/or confounding. Third, the Saskatchewan Ministry of Health's health services databases used in our study contain reliable information on which to base an assessment of the effect of drug exposure in pregnancy upon perinatal outcomes.[13] A unique identifier for all health services covered by the provincial government can be reliably used to link different databases, including prescription drug data, hospital service data, physician services data, vital statistics, and other related health information. This rich health service database is population based, and covers the majority of the population in the Province of Saskatchewan. In addition, the Provincial Drug Formulary is quite comprehensive. The drugs listed in the formulary are under continuous review: additional drugs are added if they meet the standards of professional expert committees that advise the provincial government in this regard. Prescription drug data were captured for drugs listed in the formulary and dispensed to eligible subjects. The registry was updated on a daily basis and captured most prescription drugs in the province. The hospital and physician service data used standard international coding systems. Validation, mostly by hospital chart review, has been built into several studies using such data. A recent study found high sensitivity and specificity, as well as high positive and negative predictive values, for the diagnosis of depression based on ICD–9 codes alone, which supports the validity of Saskatchewan's health services databases.[14]

It should be acknowledged that this study shares certain limitations found in most of the observational studies of drug safety. Some potentially important variables, such as non-prescription drugs and lifestyle factors, such as tobacco and alcohol use, which may affect perinatal outcomes, were not available and could not be controlled for in the analysis. As a result, residual confounding may still exist even after extensive adjustment by regression analysis and data restriction. Administrative data is prone to a certain degree of coding errors, and it is impossible to get compliance data in this historical cohort study; however, the misclassification of outcomes caused by either coding error or non-compliance may have occurred randomly, which would tend to attenuate the observed effects.[15] Because of confidentiality issues, we have relied on pre-coded variables (e.g. preterm birth as defined as <37 complete weeks of gestation) instead of the original records (e.g. gestational age). Our study found that the rate of preterm birth was actually higher in the methyldopa group than in the beta-blockers group. This finding has not been reported previously and is unexpected. Although we cannot exclude the possibility that this unexpected finding is a false positive, an examination of clinically more relevant outcomes (such as very preterm or gestational age < 32 weeks) could help us to assess the importance of such an unexpected result. Our study covered a period of 25 years, and during this time the patterns of beta-blocker use have changed, with a substantially increased use of labetalol compared with other beta-blockers in recent years (data available upon request). The research file we received comprised mother–baby pairs, and there is no unique identifier in the database for the same mothers, who may have delivered more than one baby; however, supplemental analyses restricted to first pregnancies yielded similar, although statistically less stable, results because of the reduced number of subjects after restriction. Finally, although the overall study sample size was large, the numbers of subjects diminished rapidly in subgroup analyses (e.g. only 628 for chronic hypertension).

Interpretation

The interpretation of results from observational studies is challenging, because the decision for treatment with a particular drug cannot be ascertained from these data. Given the relatively low potency of methyldopa as an antihypertensive agent, women treated by methyldopa may represent a group with less severe hypertension. Although there is no direct measure of the severity of disease, such as blood pressure values, in the database, some of the available variables, such as the initiation of treatment, may be able to serve as a ‘surrogate’ measure of severity. For example, for women with chronic hypertension, who began treatment prior to pregnancy, their symptoms may be more severe than in those who started treatment in the third trimester, because the severe cases may need to be treated earlier and for longer. The fact that odds ratios were not decreased, but were instead increased (from 1.77 to 1.95 for SGA < 10th percentile, from 1.67 to 2.17 for SGA < 3rd percentile, and from 2.10 to 2.17 for hospitalisation) after further adjustment for the start of treatment suggest that our study results may be valid, despite all of the limitations. Our study findings are also consistent with the results from a systematic review involving 13 clinical trials.[5] Meta-analyses based on randomised controlled trials would be superior than observational studies in general; however, studies from a single source of population, which would be more homogeneous than meta-analysis that combined data from different study populations, are more likely to produce consistent results. In light of the adverse effect of beta-blockers on fetal circulation demonstrated by laboratory investigations in animal models and human pregnancies, we believe that our study findings are valid.[10-12]

Conclusion

Treatment of hypertensive disorders of pregnancy with antihypertensive drugs, especially for non-severe ones (i.e. blood pressure < 160/110 mmHg), remains controversial.[16-18] Our study adds to previous findings on the association between antihypertensive medications and adverse perinatal outcomes, with an emphasis on the potential adverse impacts of beta-blockers in treating chronic hypertension during pregnancy. Future studies with larger samples should examine the effect of individual beta-blockers.

Disclose of interests

None to declare.

Contribution to authorship

RHX, DK, DM, MCW, KN, and SWW contributed and designed the study; DK, MCW, and SWW contributed to the acquisition of funding and data. YG performed the statistical analysis. YG, RHX, DM, KN, and SWW contributed to collating and interpreting the results. This article was drafted by RHX and SWW, and RHX, DK, DM, KN, and SWW revised it critically for important intellectual content.

Details of ethics approval

This study was approved by the Ottawa Hospital Research Ethics Board (2008471-01H, 20 October 2011). The research file was acquired from the Saskatchewan Ministry of Health's routine health services data, with no direct contact with the women included in the study.

Funding

This study was supported, in part, by a grant from the Ontario Ministry of Health and Long-term Care through its Drug Innovation grant (grant 2008-007) and a grant from the Canadian Institutes for Health Research (CIHR; grant MOP 86537).

Acknowledgements

S.W.W. is a recipient of a mid-career award from CIHR's Institute for Gender – Ontario Women's Health Council. D.K. is the Natural Sciences and Engineering Research Council of Canada Chair in Risk Science at the University of Ottawa. M.C.W. is supported by a University of Ottawa Tier 1 Chair in Perinatal Epidemiology. The authors thank Dr Xi-kuan Chen for his contribution at the early stage of this project. This study was based on de-identified data provided by the Saskatchewan Ministry of Health. The interpretation and conclusions contained herein do not necessarily represent those of the Government of Saskatchewan or the Saskatchewan Ministry of Health. We thank Dr Monica Taljaard of the Ottawa Hospital Research Institute for assistance with the statistical analysis of the data.

Ancillary