Objective To identify risk factors for hepatitis C virus (HCV) infection among pregnant women seeking antenatal care in tertiary care hospitals of Karachi, Pakistan.
Methods We enrolled 119 cases and 238 controls. Cases were enzyme-linked immunosorbent assay (ELISA III) positive pregnant women for antibodies to HCV; controls were anti-HCV ELISA negative pregnant women.
Results The mean age of study subjects was 26 years (SD 5) ranging from 15 to 50 years. The mean number of pregnancies for cases was 4 (SD 3) and for controls was 3 (SD 2). Among cases an average number of injections in any month was 40%, history of hospitalization was 61% and household contact with jaundice or hepatitis was 35%. In the final multivariable logistic regression model, five or more gestations (aOR = 1.99; 95% CI = 1.08–3.33), ≥1 injection (aOR = 2.33; 95% CI = 1.38–3.91) per month, hospitalization (aOR = 1.78; 95% CI = 1.01–2.99) and household contact with jaundice/hepatitis (aOR = 3.32; 95% CI = 1.89–5.83) were independently associated with HCV.
Conclusion Iatrogenic exposure (health care injections, hospitalizations and gestations) is the major risk factor for transmission of HCV among pregnant women.
Étude cas-témoins des facteurs de risque associés à l’infection à l’hépatite C chez les femmes enceintes dans les hôpitaux de Karachi
Objectif identifier les facteurs de risque pour l’infection par le VHC chez les femmes enceintes recherchant des soins prénataux dans les services dans les hôpitaux de soins tertiaires de Karachi, au Pakistan.
Méthodes Nous avons recruté 119 cas et 238 contrôles. Les cas étaient des femmes enceintes positives au test ELISA III pour les anticorps anti VHC, les contrôles étaient des femmes enceintes négatives pour ce test.
Résultats L’âge moyen des sujets étudiés était de 26 ans (écart type, SD=5), allant de 15 à 50 ans. Le nombre moyen de grossesses par cas était de 4 (SD=3) et pour les contrôles 3 (SD=2). Chez les cas le pourcentage du nombre moyen d’injections par mois était de 40%, pour un historique d’hospitalisation 61% et des ménages en contact avec la jaunisse ou l’hépatite 35%. Dans le modèle final de régression logistique multivariée, 5 grossesses ou plus (rapport de cotes ajusté«aOR» = 1,99; IC95%: 1,08, 3,33), plus d’1 injection par mois (aOR = 2,33, IC95%: 1,38; 3,91), l’hospitalisation (aOR = 1,78; IC95% = 1,01 - 2,99) et le contact des ménages avec la jaunisse ou l’hépatite (aOR = 3,32; IC95%: 1,89- 5,83) étaient des facteurs indépendamment associés au VHC.
Conclusion L’exposition iatrogène (injections pour soins de santé, hospitalisations et grossesses) est le facteur de risque majeur pour la transmission du VHC chez les femmes enceintes.
Estudio caso-control de los factores de riesgo asociados con la infección por Hepatitis C en mujeres embarazadas en hospitales de Karachi
Objetivo Identificar los factores de riesgo de la infección por VHC en mujeres que buscaban cuidado antenatal en hospitales terciarios de Karachi, Paquistán.
Métodos Se incluyeron 119 casos y 238 controles. Los casos eran mujeres embarazadas, positivas mediante prueba de ELISA III, para anticuerpos anti HCV; los controles eran mujeres con un ELISA anti VHC negativo.
Resultados La edad promedia de los sujetos del estudio era de 26 años (DS 5) con edades comprendidas entre los 15 y 50 años. El número promedio de embarazos para los casos era de 4 (DS 3) y para los controles era de 3 (DS 2). El número promedio de inyecciones entre los casos en cualquier mes era del 40%, el historial de hospitalización del 61% y los contactos familiares con ictericia o hepatitis era del 35%. En el modelo final de regresión logística multivariable estaban independientemente asociados con VHC el haber tenido 5 o más gestaciones (OR promedio=1.99; 95% IC= 1.08, 3.33), ÿ inyección por mes(OR promedi o=2.33; 95% IC=1.38, 3.91), la hospitalización (ORpromedio=1.78; 95% IC =1.01, 2.99), o tener contactos familiares con ictericia o hepatitis (ORpromedio=3.32; 95% IC=1.89, 5.83).
Conclusión La exposición iatrogénica (inyecciones en servicios sanitarios, hospitalizaciones y gestaciones) es un factor de riesgo importante para la transmisión de VHC entre mujeres embarazadas.
Hepatitis C virus (HCV) infection is major public health problem worldwide (HCV prevalence: 3%), but especially in the Eastern Mediterranean region (4.6%; WHO 2000). Although no nationally representative estimates are available, prevalence of HCV in the Pakistani population ranged from 4.5% to 6.5% (Luby et al. 1997; Muhammad & Jan 2005). In patients on hemodialysis the prevalence is 23.0% (Khokhar et al. 2005). This high prevalence of infection is translating into huge burden of chronic liver disease, cirrhosis and hepatocellular carcinoma (Giannini & Brechot 2003). These chronic sequelae of infection extend beyond ill health to economic and social disruption by loss of young productive workforce and caregivers at homes (Ahmad 2004). Further, treatment is expensive (Shepard et al. 2005), so population-based primary prevention approaches are necessary.
Studies to investigate the risk factors of HCV among males throughout Pakistan suggest that health care injections use is major driving factor of the HCV epidemic in the country. Other risk factors include shaving by barbers and blood transfusions and dental treatment (Bari et al. 2001; Akhtar et al. 2004; Khokhar et al. 2005). Most of these factors are related to health care use. Women use more health care than men due to antenatal care and child birth, which may result in surgical procedures, hospitalization, blood transfusion or unsafe medical injections that expose women to iatrogenic exposure to HCV. In urban subdivisions of the district of Hyderabad (Pakistan), 96% of women received prenatal care. Of these, 83% were attended by skilled providers in hospital (Jehan et al. 2007). Practices that particularly increase women’s risk of acquiring infection have not been investigated in Pakistan. Investigation of risk factors for HCV among women in a high prevalence country will help in promoting their health by identifying and hopefully preventing these risks. With this aim, we conducted a case–control study among pregnant women in Karachi.
This study was conducted in Karachi, the world’s second most populated city with an estimated population of 18 m from various ethnic groups (CDGK 2007). Study subjects were enrolled from the outpatient and inpatient departments of the two public hospitals, Civil Hospital, Karachi and Sobhraj Hospital of Karachi, both of which provide services at low cost. Around 4000 deliveries per year take place at these hospitals. Patients from urban, suburban and rural areas belonging to lower middle and lower socioeconomic strata seek care at these hospitals. At the outpatient departments, women came for antenatal care visits and in the inpatient departments women were admitted for delivery or had already been delivered.
Our study hospitals had a policy of advising universal screening of HCV infection. This study was conducted on the subset of pregnant women visiting hospital for antenatal care who were tested for HCV. Testing was done in a hospital laboratory in collaboration with the AIDS control programme in Sindh and cost about Rs 100 (€1.3). Almost 95% of pregnant women visiting these hospitals underwent screening (N. Hossain, personal communication).
Cases and controls
Cases were women at any stage of pregnancy or within 7 days after delivery with positive serological evidence of HCV by enzyme-linked immunosorbent assay (ELISA III) who attended the study hospitals. Controls were women at any stage of pregnancy or within 7 days after delivery with negative serological evidence of HCV by ELISA III who attended the study hospitals. A case or control was excluded from the study if they were not able to respond to questions or were also positive for hepatitis B serum antigen because the risk factors are overlapping for hepatitis B and C.
We used a pre-tested structured questionnaire translated into Urdu to collect information on subjects’ age, schooling and ethnicity, the number of household members and monthly household income in Pakistani rupees. Age in completed years was self-reported. Schooling was categorized into no formal schooling, primary (1–5 years), secondary (6–10 years), intermediate and above. Number of household members was categorized into one to six members and more than six members according to the average number of household member in urban Sindh (Statistics 1998). Self-reported information on duration of marriage, past antenatal care visits, history of abortion and place of delivery was collected. Antenatal care was categorized as subjects with at least one antenatal care visit in hospital for any previous pregnancy and subjects with no antenatal care visit for any previous pregnancy. Place of delivery was categorized as home delivery (all previous deliveries at home) and hospital delivery (at least one previous delivery at hospital). Gravidity and parity were obtained from clinical records. Self-reported history of injections/infusions use ever in life and average number of injections in a month were assessed. Self-reported information on subjects’ past history of transfusion of whole blood or blood products, of hospitalization, of past dental visits, endoscopy and history of ear or nose piercing and tattooing was also obtained.
Data were double entered by data entry operators in EPI-Info version 6.04. We performed analysis using Statistical Package for the Social Sciences (spss, version 13.0). Descriptive statistics were computed for categorical variables for cases and controls by computing their frequencies for the two groups. To assess univariate associations between HCV seropositivity and potential risk factors, odds ratio (ORs) and their 95% confidence intervals (CIs) were computed by logistic regression analysis. Quartile analysis for continuous variables was performed to check the linearity of the variable. All risk factors with P < 0.20 on univariate analysis were considered for inclusion in the model. The cut-off point for confounders was 10% for the change in beta coefficients. Plausible interaction terms between independent variables were assessed after selection of the main effect model. The final model was tested for goodness of fit by using Pearson’s chi-square statistic in the sas version 9.1.3. After checking the goodness of fit of the model, adjusted ORs and their 95% CIs were used to interpret our model.
The study protocol was approved by the Aga Khan University ethical review committee and the ethics review board of Dow University of Health Sciences. Informed consent was sought from each participant.
Between October 2005 and February 2006, 5621 pregnant women presented in the study hospitals, of whom 5339 were screened (Figure 1). A total of 120 (2%) were HCV positive; one refused to participate. Thus, 119 HCV positive pregnant women agreed to participate. We excluded 53 pregnant women with concomitant HCV HBV and 102 women with HBV positive results. There were 5064 eligible controls; we enrolled 238 controls as per 1:2 ratio of cases and controls. There was no refusal among controls.
Description of study population
About 96% of cases and controls were residents of Karachi. A total of 89% (106) of cases and 90% (215) of controls were from Civil Hospital. The mean (±SD) age of cases was 27 (±6) years and of controls 25 (±5) years. A total of 50% of cases and 39% of controls never went to school and only 7% of cases and 10% of controls had more than 10 years of schooling. A total of 40% of cases spoke Urdu, 17% spoke Pushto. The mean duration of marriage was 8 years (±6) for cases and 6 years (±5) for controls. The mean number of pregnancies was 4 (±3) for cases and 3 (±2) for controls. Increased proportion of cases (27%) had more than three children than controls (18%). Approximately half of study subjects had an income >3000 Rs (>€32) (Table 1).
Table 1. Univariate analysis showing crude odds ratios and their or 95% confidence intervals for the association of various factors with hepatitis C virus (HCV) infection among pregnant women in Karachi, Pakistan 2005–2006 (cases = 119, controls = 238)
Cases, n (%)
Controls, n (%)
*Analysis done on: no. of cases = 98, no. of controls = 182, remaining were Primipara. P-value for selection is 0.2.
Age in years (mean ± SD)
Intermediate or above
Secondary 6–10 years
Primary 1–5 years
No formal schooling
Duration of marriage
In years (mean ± sd)
Number of gestation
5 and more
Income per month
Number of household members
Previous antenatal care visits
Type of delivery*
History of abortion
Place of delivery*
History of injection/infusion
Average number of injections in any month
1 or more
History of jaundice
2 or more
Household contact with jaundice or hepatitis
History of dental treatment
History of hospitalization
History of blood transfusion
In univariate analysis, factors associated with HCV seropositivity at the 5% level were age (OR = 1.07; 95% CI = 1.02–1.11), duration of marriage (OR = 1.05; 95% CI = 1.01–1.09), a single episode of jaundice (OR = 2.09; 95% CI = 1.20–3.66), two or more episodes of jaundice (OR = 4.11; 95% CI = 1.64–10.32), household contact with jaundice/hepatitis of jaundice (OR = 3.64; 95% CI = 2.13–6.20), average number of injections in any month (OR = 2.8; 95% CI = 1.71–4.57), history of hospitalization (OR = 1.61; 95% CI = 1.03–2.51) and history of dental treatment (OR = 1.66; 95% CI = 1.01–2.70).
Parity, income, number of household members, previous antenatal care visits and history of abortions were also associated with HCV seropositivity at the 20% level. These variables were considered for inclusion in the multivariable model (Table 2).
Table 2. The multivariable logistic regression model of risk factors associated with hepatitis C virus (HCV) infection among pregnant women in Karachi, Pakistan 2005–2006 (cases = 119, controls = 238)
Cases, n (%)
Controls, n (%)
Number of gestation
5 and more
Average number of injections in a month
1 or more
History of hospitalization
Household contact with jaundice or hepatitis
Multivariable logistic regression model
Multivariable logistic regression revealed that the number of pregnancies, average number of injections in a month, household members with jaundice/hepatitis and number of hospitalization in the past were significantly associated with being anti-HCV positive (Table 2). Pregnancy showed a dose–response relationship with HCV infection. After adjusting for the effect of other variables in the model and by taking one or two pregnancies as a reference, cases were more likely to have three to four pregnancies (OR = 1.25; 95% CI = 0.70–2.24) and five or more pregnancies (OR = 1.99; 95% CI = 1.08–3.33) than controls. Taking no injection as a reference category, cases were more likely to receive on average at least one injection (OR = 2.40; 95% CI = 1.38–3.91) per month. Cases were more likely to be ever hospitalized than controls (OR = 1.78; 95% CI = 1.01–2.99). Household contact with jaundice/hepatitis was also associated with HCV, as cases were more likely to have household members who had jaundice/hepatitis previously (OR = 3.32; 95% CI = 1.89–5.83). The Pearson goodness of fit test for the final model was (χ2 = 27.89, P > 0.05).
Our study identified the association of HCV seropositivity with the number of pregnancies, therapeutic injections, past history of hospitalization and household contact with jaundice among pregnant women seeking treatment in hospitals of Karachi. The results indicate that iatrogenic transmission is major contributor of HCV infection among women in Karachi. This calls for strengthening patient safety at health care facilities, which requires comprehensive infection control steps, especially use of sterilized instruments and use of a new syringe for every injection.
The association of therapeutic injections with HCV seropositivity has been reported consistently in studies of the male population in Pakistan (Luby et al. 1997; Khan et al. 2000; Bari et al. 2001) and elsewhere, such as Taiwan (Wang et al. 1998, 2002; Sun et al. 2001), China (Ho et al. 1997) and Egypt (Habib et al. 2001). However, risk factors for HCV infections among women were not known in Pakistan. Transmission of HCV and other blood-borne pathogens is efficient if injection equipment is contaminated with blood of infected patients. Studies in Pakistan have reported that injection overuse is very common in Pakistan and most injections are provided with previously used equipment (Janjua et al. 2005). Patients are unaware of the risks associated with injections and hold exaggerated beliefs in their efficacy; practitioners, on the other hand, believe that patients want injections and have economic incentives to use them. Hence, a large proportion of health care visits results in injection prescription (Altaf et al. 2004; Janjua et al. 2006). Women are at higher risk of receiving injections because of their greater health care needs. Thus, reducing the reuse of injection equipment and overuse of unnecessary injections would be a major step towards reducing transmission of HCV among women. The Government of Pakistan has recently launched a hepatitis prevention and control programme. Our results emphasize the need for more concerted efforts towards behaviour change among both patients and health care providers.
Number of pregnancies and hospitalization provide further evidence that poor infection control practices in Karachi put women at higher risk of HCV infection. The association may be explained by the greater probability of exposure of the pregnant women to infectious environment. Pregnancy has a dose–response relationship with HCV infection, strengthening evidence towards increasing risk of infection with increasing inpatient care, especially related to pregnancy. A study of male volunteer blood donors in Karachi has shown that cases were more likely than controls to have reported past hospitalization (Akhtar et al. 2004). Studies from United States have also reported increased risk of HCV infection with hospitalization (Hyams et al. 1989; Batty et al. 2001). Our data show that 54% of women had a history of surgical interventions with more cases than controls having had surgery. Previous studies in Pakistan, Poland, Italy and Turkey suggest that previous surgeries had a significant association with HCV infection (Mele et al. 2000; Chlabicz et al. 2004; Jaffery et al. 2005; Karaca et al. 2006). Our results show that caesarean sections and abortions were not risks for HCV. However, we did not have information for general and gynaecological surgeries. Moreover, pregnant women are more likely to receive therapeutic injections and interventions such as episiotomy (which is performed routinely in primipara women in these hospitals) during their normal deliveries. Hospitalization as a risk factor flags the need for the universal precautions in health care to prevent the nosocomial exposure to HCV. Data from other studies substantiate the findings of poor infection control and non-adherence to universal precautions at first-level care facilities. Furthermore, health care workers are not trained in universal precautions (Janjua et al. 2007). National regulatory requirements for infection control are badly needed.
Household contact with a member who had jaundice/hepatitis was associated with elevated adjusted odds of HCV seropositivity among cases compared with controls. Other studies confirm these findings (al-Nasser 1992; Chang et al. 1994; Neal et al. 1994; Mohamed et al. 2005). The possible explanation could be by infectious blood or saliva, sharing a needle, razor and toothbrush within the household. Another study in Karachi revealed that sharing of toothbrushes was significantly associated with HCV seropositivity among household contacts of HCV positive patients (Akhtar et al. 2002). Some studies have reported a significant amount of HCV RNA in saliva in a substantial number of patients with chronic hepatitis (Couzigou et al. 1993; Pastore et al. 2006). Household clustering may have been reported because of similar behaviours and exposure to same external exposure sources such as therapeutic injections (Rao et al. 2002), although not assessed in our study it could be another explanation for interfamilial clustering. Inferences about the causal relationship should be drawn with care because of the case–control design. The implications are relevant to the members of high risk group and to the household members of already infected HCV patients to reduce and to prevent the spread of HCV.
This is the first case–control study of HCV risk factors among women in Pakistan and included exploration of pregnancy-specific factors. However, case–control studies have an inherent limitation for establishing direction of causal inference, especially for HCV which is a chronic and mostly asymptomatic disease (CDC Staff Members 1998). Even if cases were diagnosed for the first time at enrolment in our study, they may have been harbouring the infection for years. While health care needs may increase because of infections, need for surgery and gestations do not get extenuated with HCV infections. Therefore not all increase in health care need associated with HCV infection can be attributed to infection.
Ideally in a case–control study using hospitalized cases should identify the reference population that is the source of the cases so that this reference population can be sampled to select controls. But our study hospitals do not have well-defined catchment areas. Since we selected controls from the same hospitals presenting for same reasons as cases, both were subject to the same selection factors.
Our study participants represent the subset of pregnant women who either took antenatal care or delivered at hospitals; they were not representative of all pregnant women. In Karachi, 70% of women deliver in hospitals (National Institute of Population Studies 2007). They could be more aware or could be of better socioeconomic status. Thus, we caution about generalizability of the study.
There is always a potential of recall bias and imperfect recall in case–control design. We attempted to minimize this problem by enrolling hospital controls to have comparable recall. Furthermore, given the long duration of the asymptomatic phase of HCV infection, cases were as likely as controls to recall exposures potentially associated with HCV infection.
Household contact with jaundice/hepatitis has been identified as a risk factor for HCV infection but it was not feasible for us to ascertain the anti-HCV status of the study subjects’ household.
There is a potential of bias due to self-report; we used an interview approach and many of the risk factor questions are affected by the interpretation of the person being interviewed. Also, our study participants were aware of their HCV status and this could have affected their responses. However, any bias that might have occurred must be non-differential in cases and controls.
Conclusion and recommendations
This study showed that iatrogenic exposure (health care injections, hospitalizations and pregnancies) are the major risk factors for transmission of HCV among pregnant women. This calls for strengthening the prevention aspect of the hepatitis control programme to focus on behaviour change for reducing injection reuse and overuse. Further steps are required to enhance infection control practices at health care facilities. This could start with establishing infection control committees in hospitals, providing training to health care workers and measures to enforce adherence to universal precautions.
We thank the administrations of the Ethics Review Board of Dow University of Health Sciences and the Civil and Sobhraj Hospitals in Karachi for their cooperation during data collection.