Factors responsible for mother-to-child HIV transmission in Ontario, Canada, 1996-2008.

Lu, Dayu ; Liu, Juan ; Samson, Lindy 等

Mother-to-child (MTC) transmission is the primary mode by which infants become infected with human immunodeficiency virus (HIV) in Canada. Transmission may occur in utero, during labour and delivery or via breastfeeding. Without antiretroviral prophylaxis (ARP), 15-40% of children born to HIV-infected mothers become infected. (1-3) In 1994, the AIDS Clinical Trials Group study (ACTG 076) found that zidovudine (ZDV) given to the mother during pregnancy, during delivery and to the infant could reduce the risk of MTC HIV transmission from 25.5% to 8.3%. (3) With combination antiretroviral therapy (cART) and optimal pre- and perinatal care, risk of transmission can be reduced to less than 2%. (4-6)

Identifying HIV-infected pregnant women is critical to prevent MTC HIV transmission. In 1999, Ontario adopted a new recommendation to offer HIV counseling and voluntary HIV testing to all pregnant women in the province. Prenatal HIV testing subsequently increased from 33.2% in early 1999 to 94.6% in 2008. (7) Despite this increased testing, MTC transmission of HIV continued to occur in Ontario, with 37 documented cases reported from 1999 to 2008. The purpose of our study was to investigate the factors associated with MTC transmission of HIV in Ontario in the cART era, in particular to examine factors which could potentially reduce such transmission. Specifically, we wished to 1) characterize the correlates of late HIV diagnosis among HIV-infected mothers and 2) determine the modifiable factors responsible for transmission, including prenatal HIV testing, use of ARP and breast-feeding.

METHODS

Our study had the following components: to characterize late HIV diagnosis in mothers (Component 1) and to determine specific factors responsible for MTC transmission, including HIV testing, ARP and breast-feeding (Component 2). The Research Ethics Board at the University of Toronto and the three participating hospitals approved the study.

Component 1: Characterizing late HIV diagnosis among HIV-infected mothers

Data pertaining to Ontario was obtained from the Canadian Perinatal HIV Surveillance Program (CPHSP). Since 1991, the CPHSP has conducted national surveillance of infants and children born to HIV-infected women using data from public health units and pediatric care centres across Canada. CPHSP recently published a detailed report of the results of their surveillance. (6) We identified 645 infants born in Ontario to HIV-infected mothers from 1996 to 2008. Women were considered to have early HIV diagnosis if: 1) the mother's date of HIV diagnosis was prior to the infant's date of birth, or 2) the mother received antiretroviral prophylaxis during pregnancy for those with unknown date of HIV diagnosis; otherwise they were considered to have late HIV diagnosis. We selected this cut-off on the basis that the second (intrapartum) and third (postpartum) components of ARP could be implemented if HIV were diagnosed prior to delivery. While this might not be as effective in preventing MTC transmission as the full three-component ARP regimen, it provided an easily defined cut-off for intervention and constituted a minimum, if not ideal, time of HIV diagnosis. We classified completeness of ARP as follows: a) complete ARP when the mother received combination antiretroviral therapy during pregnancy and intravenous ZDV during labour and the infant received a six-week course of ZDV (with or without other ARP drugs); b) incomplete ARP when appropriate ARP was not administered during one or two of these periods; c) no ARP when no ARP was given. Maternal exposure category was classified according to a mutually exclusive risk hierarchy based on the most likely source of HIV infection as follows: injection drug user (IDU), born in an HIV-endemic country, heterosexual contact other than in an HIV-endemic country, blood transfusion, clotting factors, and perinatal acquisition.

Component 2: Characterizing the modifiable factors responsible for MTC transmission

Medical charts of infants with perinatal HIV infection born from 1999 to 2008 were reviewed to obtain information on laboratory results, mother's immigration status and social history, prenatal care including ARP, adherence and breast-feeding history. Infants were classified as being HIV-infected if they had a positive HIV PCR test prior to 18 months of age or a positive HIV antibody test at age 18 months of age or later. The review of the social history of the mother was assessed for illicit drug use, commercial sex work and mental illness. Details of ARP uptake were assessed by the type and timing of drug regimens administered during pregnancy, at delivery and to the infant. We also interviewed doctors, nurses and social workers who had cared for these children for cases with incomplete charts. These interviews focused only on the areas where additional data were required, were carried out for approximately 10 of the 35 cases, and generally lasted 15-20 minutes.

Statistical analyses

Summary statistics were used to describe the characteristics of HIV-infected mothers. Differences in proportions were assessed using (2) tests and Fisher's Exact Test when expected cell size was less than five. We performed multivariate logistic regression analysis to identify factors associated with late HIV diagnosis of the mother; we obtained odds ratios and 95% confidence intervals for each independent variable. The earliest period was selected as the referent for the time period variable.

However, we observed strong collinearity between race/ethnicity and exposure category and, therefore, the two variables could not be included in the same model. In fact, 97% of women from the HIV-endemic category were Black and 96% of Black women were classified as HIV-endemic. Thus, the two categories capture essentially the same women. However, because we were interested in the comparisons with other races/ethnicities and other exposure categories, we analyzed the data using two independent models.

We used SAS Version 9.2 (SAS Institute, Cary, NC, USA) for significance testing and logistic regression.

RESULTS

Study population

From 1996 to 2008, 652 infants were born in Ontario to HIV-infected mothers, of whom 50 (7.7%) were HIV-infected and 595 (91.3%) were not infected; for 7 (1.1%), the HIV status of the infant was unknown. Of those with unknown HIV status, 2 died before HIV status could be determined, 1 moved away from Canada, 1 is pending and the remaining 3 were lost to follow-up. The characteristics of the women (n=645) for whom HIV status of their infants was known are presented in Table 1. Most women were either Black (63.3%) or White (26.3%), and were born in either sub-Saharan Africa (52.4%) or Canada (29.6%). Sixty-five percent were born in an HIV-endemic country and 26.3% were otherwise infected heterosexually. The number of infants born to HIV-infected mothers increased from 23 per year in 1996-1998 to 72 per year in 2005-2008. The proportion of HIV-infected infants decreased from 20.0% in the former period to 2.8% in the latter period. Compared to women whose infants were not infected, mothers of HIV-infected infants were less likely to have been prescribed ARP during pregnancy (22.0% vs. 96.8%, p<0.0001) and more likely to have had a vaginal delivery (78.6% vs. 55.6%, p=0.011).

Component 1: Characterizing late HIV diagnosis among HIV-infected mothers

Among the 645 HIV-infected women, 529 (82.0%) women were diagnosed prior to the birth of the infant (including women without an HIV diagnosis date but who received ARP during pregnancy) while 85 (13.2%) had late HIV diagnosis. We were unable to determine the timing of HIV diagnosis for 31 (4.8%) women. The proportion of women who had late HIV diagnosis significantly decreased over the study period: 28.6% in 19961998, 23.5% in 1999-2001, 14.8% in 2002-2004 and 5.7% in 2005-2008 ([chi square] for trend, p<0.0001). A higher proportion of women with injection drug use as a risk factor were diagnosed late compared to women who were born in HIV-endemic countries and women categorized as infected heterosexually (20.4%, 12.8%, and 11.0%, respectively), though the differences were not statistically significant. There was no difference in late diagnosis by race/ethnicity (White 14.8%, Black 12.3%, Asian 16.0% and other 14.7%).

The results of the two regression models are presented in Table 2. In Model 1, late diagnosis was more frequent in non-White racial/ethnic groups compared to White but these differences were not statistically significant. Late diagnosis decreased in later study periods and was significantly lower in the last two periods. With respect to Model 2, late HIV diagnosis did not appear to vary by the mother's exposure category. After adjusting for race/ethnicity, women were less likely to have a late HIV diagnosis in the later years compared with 1996-1998, as in Model 1.

Component 2: Characterizing the modifiable factors responsible for MTC transmission

Thirty-five confirmed cases of MTC transmission of HIV were identified in Ontario from 1999 to 2008. This included 28 (80%) cases from Toronto, 4 (11%) from Hamilton and 3 (9%) from Ottawa.

Potential underlying reasons for vertical HIV transmission in these cases by maternal exposure category are presented in Figure 1 and Table 3. Seventy-seven percent (27/35) of these mothers were diagnosed with HIV at or after delivery, 6 prior to conception and 2 during pregnancy. Seventy-one percent (25/35) were from an HIV-endemic country, 2 were IDU, 7 were infected through heterosexual contact and 1 had unknown risk factors. Thirty-four percent (11/35) of these mothers breast-fed their infants.

[FIGURE 1 OMITTED]

Of the six mothers diagnosed with HIV prior to conception, all received either no or incomplete ARP. Of these, 4 refused to take ARP or were non-compliant and 2 failed to inform their care provider of their HIV-positive status.

Of the 2 mothers diagnosed during pregnancy, 1 was unaware of her HIV-positive status until two weeks prior to delivery and did not receive antenatal or intrapartum ARP, and 1 was diagnosed at 16 weeks gestation but initiated ARP only nine days prior to delivery.

Among the 27 mothers diagnosed at or after delivery, 5 had a prenatal HIV test, 13 no HIV test, and for 9 the history of prenatal HIV testing was unknown. Among the 5 mothers with a prenatal HIV test, 1 knew her positive HIV status at the time of delivery, 2 were tested 1-3 weeks prior to delivery but did not know their result, and 2, who had HIV-infected partners, had a negative HIV prenatal test during pregnancy (1 tested HIV-negative three months prior to delivery and for the other, the time of the HIV-negative test was unknown) and were infected subsequently. The reasons for no or unknown prenatal HIV test were: not offered (6), offered but refused (2), no prenatal care (3), denied testing history (4), and offered but not done (1).

DISCUSSION

From 1996 to 2008, 652 infants were born in Ontario to HIV-infected mothers, of whom 50 became infected. As in previous studies, our results revealed that infants with more complete antiretroviral prophylaxis were less likely to become infected (complete ARP 1.2%, incomplete ARP 2.4%, and no ARP 67.2%). (3, 6, 8-11) For Canada as a whole from 1997 to 2010, the Canadian Perinatal HIV Surveillance Network (CPHSP) found a vertical transmission rate of 1.0% among infants born to mothers who received cART, 1.6% for women who received mono-therapy or dual-therapy with nucleoside reverse transcriptase inhibitors, and 16.4% for women who received no ARP antenatally. For mothers who received antenatal cART at least four weeks before delivery, the MTC transmission rate was 0.4% compared to 9.0% when started later than four weeks before delivery. (6) In the UK and Ireland, the transmission rate was 0.8% in women receiving cART at least two weeks before delivery. (9) A key distinction between the CPHSP data and ours is that the data in the former are restricted to infants of HIV-infected women referred to a CPHSP participating site either before or during pregnancy or within three months following delivery (prospective cohort), whereas our study also included children diagnosed later. As a result, our study includes a higher proportion of cases of vertical transmission and, due to selective recruitment of HIV-positive infants, cannot be used to determine MTC transmission rates as the true denominator is not known.

The increase in overall prenatal HIV test uptake in Ontario and the decline in proportion of pregnant women with late HIV diagnoses in 1999-2001, 2002-2004, and 2005-2008 compared with 1996-1998 confirm the overall effectiveness of the Ontario Ministry of Health and Long-Term Care policy of offering HIV testing to all pregnant women in Ontario. Nevertheless, our study confirms that challenges remain. Late maternal HIV diagnosis was found in 27 of 35 (77%) vertical HIV transmission cases from 1999 to 2008. Furthermore, a substantial proportion (~7-8%) of HIV-infected women whose infants were not infected had late HIV diagnosis. This latter group is significant from a public health perspective because, while the infants of these mothers were not infected, they could well have been. Thus, our data confirm the need to improve timely HIV testing during pregnancy to further reduce vertical HIV transmission. One solution is to emphasize the role of HIV testing in women contemplating pregnancy in Ontario and to modify testing strategies so that marginalized women who do not access routine prenatal care can still be tested during pregnancy. Our study did not examine in depth the reasons many HIV-infected pregnant women were not tested, in particular whether these were related to the women themselves, their health providers or the organization of health services. With respect to women-related issues, cultural and psychological factors may be important given the sensitivity of HIV testing and the implications of a positive HIV result. Further research on the factors responsible for women not testing would be useful in helping to reduce the barriers to testing.

A second category of missed opportunities relates to women tested antenatally for whom appropriate preventive interventions were nevertheless not implemented in a timely manner. This is exemplified by the two women in our study who were tested antenatally, but whose results were available only after delivery. This problem could readily be rectified through the use of a rapid point-of-care HIV test (or a standard HIV serologic test ordered on an emergency basis if the former option is not available) for women presenting at delivery without an HIV test on record. The Mother-Infant Rapid Intervention at Delivery (MIRIAD) trial found it acceptable and feasible to offer counseling and rapid HIV testing at the time of labour and delivery to women of unknown HIV status. (12, 13) Last, repeat testing of women at high risk of HIV acquisition later in gestation should be considered; this is illustrated by the two cases in our study who tested HIV-negative early in pregnancy and whose partners were known to be HIV-infected. In addition to serodiscordant couple scenarios, repeat testing might be appropriate for women who use intravenous drugs, those involved in the sex trade and those who have multiple sexual partners during pregnancy. The role of routine repeat testing for all women is less clear. A cost-effectiveness analysis in the US found that a second HIV test could prevent perinatal transmissions with a net savings to society when HIV incidence among women of childbearing age is [greater than or equal to] 1.2 per 1000 person-years. (14)

There are several limitations to our study. The CPHSP database does not include the mother's date of birth. Thus, we were unable to examine mother's age in our analyses. Due to loss to follow-up, the final HIV status of some infants was unknown; however, we estimate we would have missed at most only 2-3 children vertically infected with HIV in this way. We also found errors in the database, as we identified one duplicate case, one case incorrectly identified as HIV-infected, one case with incorrect birthplace of the infant, and one case which was not reported to the CPHSP; thus, there could have been other unidentified errors in the database. Furthermore, in the case review component of our study, the information about the mother's history relied primarily on notes in the infants' chart. However, we also collected such information through interviews with doctors, nurses and social workers; recall bias could have been introduced. Uptake of prenatal care, prenatal HIV testing and breast-feeding history were unknown for some cases, which could have led to inaccuracies in our results. Finally, we examined only some of the modifiable factors potentially responsible for MTC transmission; we did not, however, examine the biologic factors responsible for such transmission, such as maternal viral load and mode of delivery.

While overall HIV test uptake in Ontario has improved dramatically subsequent to the implementation of the Ontario Ministry of Health and Long-Term Care policy of universally offering HIV testing in pregnancy in Ontario, (15) significant gaps remain. The importance of offering HIV testing to all women early in gestation and to women contemplating pregnancy is paramount, as is the need for repeat HIV testing later in gestation for women at significant ongoing risk of HIV acquisition. We need to ensure all pregnant women have access to HIV testing, including those who do not (or cannot) access routine prenatal care. HIV testing should be timed to allow for early initiation of ARP. For women presenting late in gestation with unknown HIV status, rapid HIV testing is clearly warranted and could prevent some MTC transmissions.

Received: June 18, 2013

Accepted: December 3, 2013

Acknowledgements: We are indebted to the Canadian Perinatal HIV Surveillance Program (CPHSP) for providing the data that made this study possible. We thank the following persons at the participating hospitals for their assistance in the medical record reviews: Cheryl Arneson, Robyn Salter, Georgina MacDougall (Hospital for Sick Children in Toronto), Jennifer Bowes (Children's Hospital of Eastern Ontario in Ottawa).

Conflict of Interest: None to declare.

REFERENCES

(1.) Newell ML, Peckham C. Vertical transmission of HIV infection. Acta Paediatr 1994;400(suppl):43-45.

(2.) Mofenson L. Epidemiology and determinants of vertical HIV transmission. Semin Pediatr Infect Dis 1994;5:252-65.

(3.) Connor EM, Sperling RS, Gelber R, Kiselev P, Scott G, O'Sullivan MJ, et al. Reduction of maternal-infant transmission of HIV-1 with zidovudine treatment. N Engl J Med 1994;331:1173-80.

(4.) European Collaborative Study. Mother-to-child transmission of HIV infection in the era of highly active antiretroviral therapy. Clin Infect Dis 2005;40(3):458-65.

(5.) Cooper ER, Charurat M, Mofenson L, Hanson IC, Pitt J, Diaz C, et al. Combination antiretroviral strategies for the treatment of pregnant HIV-1infected women and prevention of perinatal HIV-1 transmission. J Acquir Immune Defic Syndr 2002;29(5):484-94.

(6.) Forbes JC, Alimenti AM, Singer J, Brophy JC, Bitnun A, Samson LM, et al. A national review of vertical HIV transmission. AIDS 2012;26(6):757-63.

(7.) Remis RS, Swantee C, Liu J. Report on HIV/AIDS in Ontario 2008. Dalla Lana School of Public Health, University of Toronto. April 2010. Available at: http://www.ohemu.utoronto.ca/doc/2011/PHERO2008_report_final_ rev_Sept2010.pdf (Accessed September 20, 2012).

(8.) Forbes JC, Money DM, Remple PV, Burdge DR. Effect of antiretroviral use on HIV vertical transmission rate and injection drug use on adherence in British Columbia, Canada. Can J Infect Dis 2000;11:46B (Abstract 246P).

(9.) Townsend C, Cortina-Borja M, Peckham C, de Ruiter A, Lyall H, Tookey PA. Low rates of mother-to-child transmission of HIV following effective pregnancy interventions in the United Kingdom and Ireland, 2000-2006. AIDS 2008;22:973-81.

(10.) McDonald A, Zurynski Y, Wand H, Giles ML, Elliott EJ, Ziegler JB, et al. Perinatal exposure to HIV among children born in Australia, 1982-2006. Med J Aust 2009;190:416-20.

(11.) Birkhead G, Pulver W, Warren B, Hackel S, Rodriguez D, Smith L. Acquiring human immunodeficiency virus during pregnancy and mother-to-child transmission in New York: 2002-2006. Obstet Gynecol 2010;115:1247-55.

(12.) Bulterys M, Jamieson DJ, O'Sullivan MJ, Cohen MH, Maupin R, Nesheim S, et al. Rapid HIV-1 testing during labor: A multicentre study. JAMA 2004;292:219-23.

(13.) Keenan-Lindsay L, Yudin MH, Boucher M, Cohen HR, Gruslin A, MacKinnon CJ, et al. HIV screening in pregnancy. J Obstet Gynaecol Can 2006;28(12):1103-07.

(14.) Sansom SL, Jamieson DJ, Farnham PG, Bulterys M, Fowler MG. Human immunodeficiency virus retesting during pregnancy: Costs and effectiveness in preventing perinatal transmission. Obstet Gynecol 2003;102(4):782-90.

(15.) Remis RS, Merid F, Palmer RWH, Whittingham E, King SM, Danson NS, et al. High uptake of HIV testing in pregnant women in Ontario, Canada. PLoS One 2012;7(11):e48077.

La traduction du resume se trouve a la fin de l'article. Can j Public Health 2014;105(1):e15-e20.

Dayu Lu, mph, [1] Juan Liu, MSc, [1] Lindy Samson, MD, [2] Ari Bitnun, MD, [3] Sandra Seigel, MD, [4] Jason Brophy, MD, [2] Lynne Leonard, PhD, [5] Robert S. Remis, MD [1, 6]

Author Affiliations

[1.] HIV Epidemiologic Monitoring Unit, Dalla Lana School of Public Health, University of Toronto, Toronto, ON

[2.] Division of Infectious Diseases, Department of Pediatrics, Children's Hospital of Eastern Ontario, University of Ottawa, Ottawa, ON

[3.] Division of Infectious Diseases, Department of Pediatrics, Hospital for Sick Children, University of Toronto, Toronto, ON

[4.] Department of Pediatrics, McMaster Children's Hospital, Hamilton, ON

[5.] Department of Epidemiology and Community Medicine, University of Ottawa, Ottawa, ON

[6.] HIV Laboratory, Public Health Ontario, Etobicoke, ON

Correspondence: Dr. Robert S. Remis, Dalla Lana School of Public Health, University of Toronto, Health Sciences Building, 155 College Street, Room 512, Toronto, ON M5T 3M7, Tel: +416-946-3250, Fax: +416-978-2087, E-mail: rs.remis@utoronto.ca

Table 1. Characteristics of HIV-infected mothers in Ontario
from 1996 to 2008

                                Mother-infant
                                    pairs

                                          n
Total                           645       50
Study period ([dagger])
  1996-1998                      70       14
  1999-2001                     115       18
  2002-2004                     173       10
  2005-2008                     287        8
Region of birth of mother
  Canada                        188       10
  Sub-Saharan Africa            333       28
  Caribbean                      62        6
  Asia                           28        4
  Other                          24        1
  Unknown                        10        1
Maternal exposure category
  Injection drug user            50        2
  HIV-endemic                   393       33
  Heterosexual                  159       13
  Transfusion                     3        0
  Unknown                        40        2
Race/ethnicity
  White                         167        8
  Black                         402       35
  Asian                          30        2
  Other                          36        4
  Unknown                        10        1
Antiretroviral prophylaxis
    ([dagger])
  Complete prophylaxis          251        3
  Incomplete prophylaxis        336        8
  None                           58       39
Mode of delivery ([dagger])
  C-section, elective           228        9
  C-section, emergency           31        0
  Vaginal                       346       33
  Unknown                        40        8

                                 HIV-infected infants

                              % infected    % column *
Total                               7.8%             -
Study period ([dagger])
  1996-1998                        20.0%         28.0%
  1999-2001                        15.7%         36.0%
  2002-2004                         5.8%         20.0%
  2005-2008                         2.8%         16.0%
Region of birth of mother
  Canada                            5.3%         20.4%
  Sub-Saharan Africa                8.4%         57.1%
  Caribbean                         9.7%         12.2%
  Asia                             14.3%          8.2%
  Other                             4.2%          2.0%
  Unknown                          10.0%
Maternal exposure category
  Injection drug user               4.0%          4.2%
  HIV-endemic                       8.4%         68.8%
  Heterosexual                      8.2%         27.1%
  Transfusion                       0.0%          0.0%
  Unknown                           5.0%
Race/ethnicity
  White                             4.8%         16.3%
  Black                             8.7%         71.4%
  Asian                             6.7%          4.1%
  Other                            11.1%          8.2%
  Unknown                          10.0%
Antiretroviral prophylaxis
    ([dagger])
  Complete prophylaxis              1.2%          6.0%
  Incomplete prophylaxis            2.4%         16.0%
  None                             67.2%         78.0%
Mode of delivery ([dagger])
  C-section, elective               3.9%         21.4%
  C-section, emergency              0.0%          0.0%
  Vaginal                           9.5%         78.6%
  Unknown                          20.0%

                              Non-infected infants

                                n     % column *
Total                          595             -
Study period ([dagger])
  1996-1998                     56          9.4%
  1999-2001                     97         16.3%
  2002-2004                    163         27.4%
  2005-2008                    279         46.9%
Region of birth of mother
  Canada                       178         30.4%
  Sub-Saharan Africa           305         52.0%
  Caribbean                     56          9.6%
  Asia                          24          4.1%
  Other                         23          3.9%
  Unknown                        9
Maternal exposure category
  Injection drug user           48          8.6%
  HIV-endemic                  360         64.6%
  Heterosexual                 146         26.2%
  Transfusion                    3         0.54%
  Unknown                       38
Race/ethnicity
  White                        159         27.1%
  Black                        367         62.6%
  Asian                         28          4.8%
  Other                         32          5.5%
  Unknown                        9
Antiretroviral prophylaxis
    ([dagger])
  Complete prophylaxis         248         41.7%
  Incomplete prophylaxis       328         55.1%
  None                          19          3.2%
Mode of delivery ([dagger])
  C-section, elective          219         38.9%
  C-section, emergency          31          5.5%
  Vaginal                      313         55.6%
  Unknown                       32

* Column percent calculated only among those cases with known
values for each variable.

([dagger]) p<0.05, excludes those with unknown value.

Table 2. Potential factors associated with late HIV diagnosis
among mothers in multivariate logistic regression

                                Late HIV
                     Total     diagnosis
                                               Adjusted Odds
                              N       %       Ratio (95% CI)
Model 1
Race/ethnicity
   White              162     24    14.8%    1.00 (Referent)
   Black              383     47    12.3%    1.02 (0.59-1.78)
   Asian               25      4    16.0%    1.45 (0.43-4.85)
   Other               34      5    14.7%    1.46 (0.49-4.39)
Study period
   1996-1998           70     20    28.6%    1.00 (Referent)
   1999-2001          115     27    23.5%    0.77 (0.39-1.52)
   2002-2004          144     19    13.2%    0.38 (0.19-0.77)
   2005-2008          275     14     5.1%    0.13 (0.06-0.28)
Model 2
Exposure category
   Injection drug      49     10    20.4%    1.00 (Referent)
     user
   HIV-endemic        374     47    12.6%    1.06 (0.47-2.37)
   Heterosexual       154     17    11.0%    0.79 (0.32-1.94)
Study period
   1996-1998           70     20    28.6%    1.00 (Referent)
   1999-2001          113     26    23.0%    0.72 (0.36-1.43)
   2002-2004          134     16    11.9%    0.33 (0.16-0.70)
   2005-2008          260     12     4.6%    0.12 (0.05-0.26)

Table 3. Potential reasons for HIV transmission by exposure category

Exposure category       N    No prenatal   No prenatal
                                care       HIV testing

HIV-endemic            25     2 (8.0%)      9 (36.0%)
Heterosexual            7     0 (0.0%)      3 (42.9%)
Injection drug user     2     0 (0.0%)      1 (50.0%)
Unknown                 1     1 (100%)      0 (0.0%)
Total                  35     3 (8.6%)     13 (37.1%)

Exposure category        No ARP     Incomplete   Breast-feeding *
                                       ARP

HIV-endemic            22 (88.0%)   3 (12.0%)        9 (36.0%)
Heterosexual           6 (85.7%)    1 (14.3%)        3 (42.9%)
Injection drug user    1 (50.0%)    1 (50.0%)        0 (0.0%)
Unknown                 0 (0.0%)     1 (100%)        0 (0.0%)
Total                  29 (82.6%)   6 (17.1%)       12 (34.3%)

* Ten mothers were diagnosed with HIV 3 months to 7 years after
delivery; one mother was diagnosed during the delivery and had
breast-fed 1.5 months; one mother was diagnosed one year prior to
delivery but did not inform care provider - her HIV status was
discovered 3 days after delivery and breast-feeding stopped.