Prevalence of HIV and Treponema pallidum infections in pregnant women in Campinas and their association with socio-demographic factors

The anonymous seroprevalence of HIV and syphilis was studied by collecting umbilical cord blood samples from 5,815 women who gave birth in Campinas' hospitals throughout a six-month period. ELISA and Western blot were used for HIV, and VORL and TPHA for Treponema pallidum screening. While maintaining the anonymity of the women, information was recorded on the hospital of origin, divided into university (public) and private hospitals, as well as on the form of payment (social security, private insurance or direct payment), age, marital status, education, employment and place of residence. Seroprevalence was 0.42 percent for HIV and 1.16 percent for syphilis. There was a significant correlation between a positive reaction to the two infections (p=0.02). After univariate and logistic regression analysis, only university hospitals were shown to be associated with seropositivity for HIV, whereas the same variable and an older age were associated with syphilis. All positive reactions were found either in public hospitals or among social security patients treated at private institutions. The conclusion was that HIV infection is becoming almost as prevalent as syphilis among this population, and affects primarily the lower socio-economic strata. This suggests that routine, voluntary HIV serology should be considered and discussed with patients during prenatal or delivery care whenever a population shows a seroprevalence close to or greater than 1 percent.


INTRODUCTION B
y mid-1991, a total of 371,802 AIDS cases had been reported worldwide; 18,113 of these were in Brazil, which ranked third in number of cases. 52he spread of the epidemic has been very rapid throughout the world, but has not followed the same model in every country.The identification of several epidemiologic patterns has contributed to a better understanding of women's participation in the spread of the disease.Epidemiologic pattern Type I, found in the United States and in some areas in Latin American, corresponds to populations in which homosexual men are the most affected group, followed by intravenous drug users and a small number of women.Type II pattern is found in Africa and the Caribbean, with predominantly heterosexual dissemination and no difference in prevalence between men and women.Pattern III refers to countries with low AIDS prevalence even among groups considered high risk. 40Such patterns, however, can change with time, giving rise to an intermediary pattern Type IIIl, found in regions where initial dissemination was among homosexuals and drug users, but where the incidence of the disease has been expanding through heterosexual transmission.41 The World Health Organization considers Latin America, including Brazil, to be part of this intermediate group.36.39According to information from the Brazilian Ministry of Health, the first male AIDS case in Brazil was reported in 1980, but the first female case was not reported until 1983.In 1984, there was one infected woman for every 124 men, but in 1993 the ratio was 1:4. 12 The evolution of AIDS distribution among Brazilian women has shown a decrease in infection through blood and blood products, but a progressive increase in infection by heterosexual contact, which today is the main mechanism for the transmission of HIV infection (Fig. I).---------------------------60 ,,~roro ,,~ro~,,~~I\j ,,~~" "~~7,,.,,~~'O

YEAR
The immediate consequence of the higher prevalence among women is the vertical transmission of the infection to their children in a proportion that varies from 12.9 to 39 percent.9.24.25,45,49In Brazil, during the first half of 1993, vertical transmission represented 87.9 percent of all pediatric AIDS cases, whereas up to 1986, 78 percent of HIV infections among children had been acquired through blood.12  Knowledge about the dimension of the epidemic is required to educate the population to adopt preventive measures, and provides the basis for sensitizing those who are politically responsible to make decisions which support control programs.This knowledge is also required for the evaluation of the impact caused by control measures.The indicator commonly used to assess the dimensions of the epidemic is the number of AIDS cases reported to health authorities.Nevertheless, such methodology affords only a late insight into the dynamics of the infection, as the numbers obtained refer to individuals who had been contaminated several years earlier.6   Prevalence studies of HI V infection constitute a more efficient method to evaluate the dimension of the epidemic and to follow its dynamics, providing global and updated information on contamination levels, which can then be applied to several population groupS.6,4HTheirrepetition in relatively stable population groups through so-called "sentinel surveillance," provides a long-term view of the evolution of the epidemic.The multiplication of these "sentinel surveillance" studies permits efficient, swift and economical screening, as well as a monitoring of the infection.20 , 47  The prevalence of the infection among women may be a good indication of the situation among the general population when non-selected groups are evaluated.The most comprehensive and practical way to reach unselected women is to perform serologic tests on patients attending prenatal clinics or at birth.29.33.3H Another reason for investigating the prevalence among delivering women is the risk of contamination for health professionals.15 .16 Prospecti ve studies of the risk faced by health professionals exposed to seropositive blood has shown a 0.3 percent rate of seroconversion after accidents with contaminated needles or cutting instruments, and no seroconversion after exposure of the mucosa. 30he professional assisting at a birth is exposed to much larger quantities of blood and other body fluids than most surgeons in non-obstetric procedures.Therefore, prevalence data on syphilis and HIV among pregnant women is also useful to highlight the need to take preventive measures.This applies to hospital management, to ensure the adequate provision of equipment, and to doctors and nurses, so as to educate them to take prot~c-tive measures.
In Brazil, the medical literature shows a prevalence ranging from 1.0 percent to 3.75 percent in prenatal patients/delivering women in specific population groups, Routine testing creates other problems, as the worldwide ethical norm for HIV infection screening establishes that the individuals to be tested should give their informed consent whenever identified by the person requesting the test.1.14.IR Prevalence studies with identified subjects who agree to be tested lead to a underestimation of up to 50 percent, as a consequence of the selfexclusion of individuals at higher risk.22.30.31.50The recommended ethical procedure for conducting random seroprevalence studies is to carry out anonymous serology, both on the part of the individual being tested and on the part the researcher, in a double-blind study.3.6.47,4RThisstudy design is considered ethical, because the blood sample becomes the object of the study, as the identity of the patients is unknown.This methodology has already been used in several countries,27.3o.33.5oand is recommended by international organizations dealing with AIDS control after careful consideration of its advantages and disadvantages.3R ,47,4R The main advantage of this method is that it provides a more accurate evaluation of prevalence.At a time when few physicians pay proper attention to HIV infection, we judged raising awareness on the actual prevalence to be a priority that requires the adherence to methodology internationally recommended for this purpose.
The development of this study in Campinas was justified because the city was ranked ninth in absolute numbers and seventeenth in AIDS incidence coefficient, 12 which placed the city in an intermediate situation among Brazilian urban centers.A parallel study on T. pal/idum infection was added with the goal of comparing the prevalence of the two diseases, considering also that syphilis serology is part of the standard prenatal routine recommended by the Ministry of Health.lo This has the additional purpose of studying lues seropositivity as a risk factor for HIV infection.

SUBJECTS AND METHOD
, A cross-sectional study was carried out between April I and September 30, 1991.Umbilical cord blood samples were collected in all deliveries occurring at the five hospitals registering the highest number of obstetric cases in Campinas (83 percent of deliveries).Three of the maternity hospitals are part of private or philanthropic organizations, and the other two are university hospitals, which make up part of SUS (Public Health System).Samples were collected during all births, without any selection criteria.Therefore, all types of patients were included, from those whose expenses were covered by SUS to those who had private health insurance or paid directly.The clinical boards of directors and the hospitals' ethics committees approved the development of the study after evaluating the proposal.
Umbilical cord blood samples (I Oml) were collected without the addition of any anticoagulant, in all births, independent of the form of del ivery (vagi nal or by Caesarean section).The patients' initials, date and time of delivery were written on adhesive labels, with specific serial numbers for each hospital printed on the same label.Samples identified only by the label were collected daily by a research assistant from the five hospitals and delivered for serologic testing to the Hemocentro -Campinas, located at the State University of Campinas (UNICAMP).
An ELISA test (Abbot HIV I EIA, manufactured by Abbott Laboratories in Brazil) was carried out on each sample, followed by Western blot (Du Pont) whenever ELISA results were positive or inconclusive.Western blot was considered positive when positive results were found for bands p 24, P 31, gp 41 or gp 120/gp 160, according to Biotech du Pont, criteria corroborated by the U.S. Food and Drug Administration.'7  Screening for T. pallidu/11 infection was conducted by VDRL quantitative test (Boehring), and all positive samples were tested for confirmation by TPHA (Biolab).
In the meantime, research assistants collected epidemiologic data from hospital records, removing any personal identification which could tie serology results to specific patients.To protect anonymity, only numerical identification was kept.
Dependent variables studied were: HIV infection and T. pal/idu/11 infection.Independent variables studied were: type of hospital (university or other); mode of payment for hospital bill (meaning how hospital and medical costs were covered); age; marital status; education; professional occupation; and city of residence.
All information was recorded on a special form.All forms without the serology results were excluded from the analysis.When serology results did not have the corresponding form with inforlllation on the independent variables, they were used only to estimate the general prevalence of the two infections (113 cases).Results of

Table 2
Number of blood samples lost to analysis, by reason serological tests were entered into a IBM-PC computer using the DBase III Plus program; statistical analysis was carried out on the SPSS-PC program.Univariate analysis was concluded using dichotomized variables through the Q-square test and, when necessary, through the Yates and Fisher correction procedures.Multiple analysis was carried out using the logistic regression technique.

T. pallidum
Ratio HIV/ n (%) n (%) T.pallidum The 0.4 percent HIV seroprevalence found among childbearing women in Campinas can be considered Jow,

DISCUSSION
Twenty-two positive and five inconclusive results were found by the ELISA method for HIY.Twenty positive and one inconclusive ELISA-samples were confirmed as positive by the Western blot test.In the non-specific test for T. pailidum, 61 reacting samples were found, with quantitative VDRL less than or equal to one-eight.The specific proof confirmed the positivity of 59 samples, with evidence that the remaining two were false-positive.
Thus, the HIV seroprevalence found in the total collection of samples was 0.4 percent or 4 per 1000, varying from 0 percent to 1.0 percent among participating hospitals.General seroprevalence for T. pailidum was 1.2 percent, varying from 0.5 percent to 2.0 percent.The highest prevalence was observed in university hospitals, and no HIV infection was detected at the one hospital which attends only private patients and those with private health insurance (Table 3).
There was a significant correlation (p<0.02) between seropositivity for syphilis and HIV.HIV infection prevalence was seven times higher in positive T. pailie/um samples (3.0 percent), than in the negative T. pailie/um group (0.4 percent) (Table 4).The univariate analysis showed a correlation of both infections with university hospital and SUS payment mode; there also was a correlation with age over 24 for T. pailie/um, and with the absence of a marital union for HIV (Table 5).No HIV seropositive samples were found among private patients or those who had private health insurance; the 21 seropositive samples were found among patients using SUS.Multiple regression showed, for both infections, a correlation with university hospital care and, for syphilis only, a correlation with older age (Table 6).During the period of sample collection there were 9,209 births.However, because of failures in sample collection, only 5,815 blood samples were obtained, corresponding to 60.13 percent of all deliveries.The number of births, samples collected, and resulting proportion by hospital are shown in Table 1.
More than 10 percent (714) of the samples were rejected for different reasons.Spontaneous formation of plasma in place of serum rendered 19 samples adequate for HIV serology, but not for T. pallidum.Consequently, 5,101 valid samples were selected for the ELISA test and 5,082 for the VDRL test (Table 2).and the growing infection rates among the .womendelivering at the university hospital in Ribeirao Preto. 23 prevalence of 0.4 to 0.6 percent in Campinas, and of 1.0 to 1.5 (if cord blood testing detects only two-thirds of maternal infections) in university hospitals should call the attention of medical authorities to the need for HIV screening in pregnant women.
The cost of HIV serology has limited the use of testing during prenatal care in Campinas.Only one university hospital offered this test, but restricted it to pregnant women who referred risk factors for HIV contamination.In view of the results of this study, this attitude may be too passive considering the expansion of the epidemic in the city.
The U.S. Public Health Department, in 1987, and the American College of Obstetricians and Gynecologists, in 1988, recommended that counseling and serologic tests should be offered in all centers assisting high-risk women or groups with increasing seroprevalence. 1n New Jersey, a HIV prevalence rate of 4.9/1 000 in umbilical cord blood samples (similar to what was found in Campinas) was the justification for offering voluntary HIV testing to all childbearing women and those intending to become pregnantY The advantage of routine prenatal screening for HIV is that pre-and posttest counseling may provide opportunities for adequate information/education and dialogue leading to genuine behavior modification.Despite the strong emotional impact of the infection diagnosis during pregnancy, there is a benefit in the closer clinical surveillance of the infected pregnant woman and her newborn, which seems to improve the prognosis for seropositive patients. 13he similarities of HIV and T. pallidum infections in young women, allied to their close prevalence, suggests the need to treat them in a similar way at the programmatic  is classified as intermediate. 51he study has shown, however, great inequality in the distribution of cases; HIV and T. pallidum prevalence in university hospitals was twice as high as in the total of the sam pIes (close to 1 and 2 percent, respectively).This classified them as having intermediate to high rates of these infections.The same ratio of one H~V to three T. pallidum infections was observed, with minor variations, in each hospital and in the entire collection of samples.
The evaluation of the prevalence found in this study should take into consideration that the precision of umbilical cord blood screening in the assessment of maternal serologic condition is questionable.Its accuracy for HIV infection has not yet been studied.As for syphilis, RAWSTROM and BROMBERG 4 4 compared serology in maternal, umbilical cord and neonate blood, using the rapid plasma reagent (RPR).These authors found 33 percent of false-negati ve results in cord blood compared to maternal infection.On the other hand, 2 percent of the cord blood samples were positive while the maternal samples were negative.The same proportion of false-negative results for syphilis, in cord blood, had been found by MILLER et a1. 37If the same margin of error were valid for HIV, the real maternal prevalence in our sample would be 0.6 and 1.8 percent for HIV and T. pallidul11, respectively.
These results are in agreement with other evidence of increased HIV infection rates among B razi Ii an women of reproductive age, such as the increasing -Prevalence of HIV and Treponema pallidum infections in pregnant women in Campinas and their association with socio-demographic factors level.For example, the populations at risk and the educational activities to prevent the two infections frequently overlap.The crucial difference is that HIV infection is non-treatable and carries a strong stigma, which justifies the requirement of informed consent prior to HIV testing, while serologic diagnosis of syphilis is routinely done without ethical conflicts.
In light of our results and the above considerations, it is suggested that HIV testing should be included in routine prenatal care in Campinas, as well as in all regions where infection rates have reached intermediate or high levels, such as Santos, Ribeirao Preto, Sao Paulo and Rio de Janeiro.21.23.26.46 Studies of pregnant women in the United States have shown that acceptance of HIV testing during the prenatal period reaches only 76 percent,4.5 and that infected women are more common among those who refuse to be screened for HIy '35  The alternative approach, selectively screening women with risk factors for HIV infection, becomes less efficient, because the higher the prevalence of HI V among the general female population, the higher the proportion of infected women not detected by risk behavior screening.22.34.5oThus, the clinical-epidemiologic evaluation of risk factors has not been effective in identifying all infected women, leaving 50 -66 percent of positive cases undiagnosed.23 .28 Therefore, several authors conclude that with prevalence reaching significant levels, routine and universal consented prenatal testing is recommended.48  The need and convenience of adopting this strategy depends on periodic reevaluation.
The intermediate position of Campinas in the scenario of the epidemic and its populational characteristics suggest that it may serve as a sentinel city for national surveillance of the disease.To this end, repeated prevalence studies among childbearing women should be carried out in Campinas at regular intervals.
The problem is that prevalence studies must be blind and universal to be accurate.The ethics of carrying out blind testing for HIV infection has been questioned because it means not identifying HIV positive women who will deliver and breastfeed their babies without being informed of their condition, or of the precautions they should take with their sexual partner and the newborn.The available alternative should be considered for ethical judgement.
The present situation in Campinas, as well as in the overwhelming majority of births in Brazil, is that women are not given the option ofHIV testing.The limiting factors are the lack of available laboratory facilities, the cost of the kits for ELISA and, particularly, of the Western Blot test.As described above, at the time the present study was carried out, only the laboratory of one of the university hospitals was agreeing to do the ELISA test for a selected group of high-risk women attending a prenatal clinic, whereas women delivering in the other hospitals in the city did not have access to HIV testing.
If HIV screening had been available for all patients delivering in Campinas at the time of the study, it would have been unethical to interrupt their routine for research purposes.This was not the case, as no routine was altered during the study period.Still, it could be criticized from an ethical point of view, that as the HIV test had been administered, even throughout a short period, those women and their families could have benefited from information about their condition.This would have required identifying the samples with the name of each woman.For this, we would have had to request the patients' consent, according to accepted ethical principles.Previous experience shows that about 25 percent of the women would not have consented, and that the prevalence of HIV infection is higher exactly in this group.21.28.31.50Thus, if we had followed the first procedure we would have failed to accomplish the purpose of the study, namely, to evaluate the real prevalence of HI V infection among women giving birth in Campinas, as those with the highest prevalence might not have been tested.
The ethics of blind surveillance testing has been the subject of profound discussions by working groups in Europe, the United States, Canada and the World Health Organization.An American Journal of Public Health editorial recently concluded that "seroprevalence studies are not only ethically permissible,but their conduct is an ethical resposibility for those responsible for public health".? This opinion was endorsed by an editor ofBioetica (ajournal published by the Brazilian Federal Council of Medicine), who reviewed the article mentioned above.8  Another problem with blind, anonymous serology is that it does not permit the detailed study of sociodemographic factors potentially associated with these infections.
Even considering such methodologic restrictions, the present data suggest that there is a link between socio-economic situation and infection, both for HIV and T. pallidul11, as all the positive cases were women treated by SUS; none of them were private patients or women with private health insurance.The correlation between HIV infection and poverty has been described by several authors.32 .42In the United States, this connection is indirectly observed in the disproportionate prevalence of the infection among blacks and Hispanics.
Consequently, it appears that HIV infection is another example of the social disadvantage of the poor.The current difficult situation of the public health system in Brazil can not be an excuse to ignore the progressive contamination of the population with HIY.In 1988, the Ministry of Health (MOH) rejected a proposal for universal testing for HIV during prenatal care.Since then, the MOH has not made a single statement concerning this issue.1I Any change in policy should be based on current data on HIV prevalence, such as that provided by this study.We hope that our results may help many women in the future if these results, and those of a later second round, are successful in convincing health authorities of the need for routine HIV screening during prenatal care.

Table 1 Proportion of valid samples to total births by hospital
AMARAL, E.; FAUNDES, A.; GONl;ALES, N.S.L. et al. -Prevalence of HIV and Treponema Sao Paulo Medical Journal/RPM 114(2): 1108-1116, 1996 pallidum infections in pregnant women in Campinas and their association with socio-demographic factors

Table 3
Seroprevalence for HIV and T.pallidum by hospital

Table 4
Correlation between seropositivity for HIV and T.pallidum

Table 6 ,
Value for factors associated with HIV and T.pallidum in multiple logistic regression

Table 5
Summary of univariate analysis according to variable categoryPrevl.= Prevalence while 1.2 percent for T. pallidum