Objectives: Tuberculosis (TB) disease adversely affects mother and child, and strategies to control TB in this group are important. The aim of this study was to analyze the epidemiology of TB in pregnancy, and to establish whether pregnancy is an independent risk factor for TB.
Methods: The United Kingdom–wide cohort study was based on the General Practitioner Research Database (GPRD), enrolling all women with pregnancies between 1996 and 2008. Incidence rates and incidence rate ratios (IRRs) of TB events during pregnancy, 6 months postpartum, and outside pregnancy were calculated and compared by Poisson regression. A nested self-controlled case series compared the risk of TB in these periods, adjusting for individual and time-bound confounders.
Measurements and Main Results: The crude TB rate for the combined pregnancy and postpartum period was 15.4 per 100,000 person-years, significantly higher than outside of pregnancy (9.1 per 100,000 person-years; P = 0.02). Adjusting for age, region, and socioeconomic status the postpartum TB risk was significantly higher than outside pregnancy (IRR, 1.95; 95% confidence interval [CI], 1.24–3.07), whereas there was no significant increase during pregnancy (IRR, 1.29; 95% CI, 0.82–2.03). These observations were confirmed in the self-controlled case series (IRR, 1.62; 95% CI, 1.01–2.58 and IRR, 1.03; 95% CI, 0.64–1.65, respectively).
Conclusions: The incidence of TB diagnosis is significantly increased postpartum. Although we did not find an increase during pregnancy, the postpartum incidence may reflect an increase during pregnancy given diagnostic, immunological and administrative delays. Clinicians’ awareness should be improved and the effectiveness of public health policy measures such as targeted screening of pregnant and postpartum women in high-risk groups should be evaluated.
Tuberculosis (TB) disease during pregnancy adversely affects mother and child, and strategies to control TB in this group are important. However, the incidence of TB in pregnant women is not known.
In a large cohort of pregnant women in the United Kingdom, we found a significant increase in the postpartum incidence of TB. Given diagnostic, immunological, and administrative delays, these results suggest an increased incidence of TB during pregnancy. The self-controlled case series found an increased TB risk postpartum, adjusting for individual confounders.
Globally, tuberculosis (TB) is a leading cause of morbidity and premature mortality (1) and one of the most important causes of death among 15- to 44-year-old women (2). In developed countries a rise in the number of pregnant patients with tuberculosis has been described, as a result of increasing numbers of cases among migrants and ethnic minority groups with younger age distribution (3).
In the United Kingdom, recent rises in TB incidence have been associated with a change in the epidemiology of TB (4), with the disease now affecting younger age groups (3) and immigrants from high-prevalence countries (4–6). TB rates in pregnant women are expected to be higher because of the different demographic composition of this group (3), and the fact that women from high-prevalence areas often have a higher total period fertility rate (7). Two local studies from large London teaching hospitals provided incidence estimates in pregnant women between 3 and 5 times higher than the respective local background rate (153 to 252 per 100,000 maternities) (8, 9). However, it remains unclear whether pregnancy increases the risk of TB or whether this observation relates to the higher occurrence of the disease in high-risk groups.
Pulmonary (10) as well as extrapulmonary TB (11) adversely affect the health of mother and child. TB in pregnancy has been shown to respond to standard treatment (9), but considerable delays in diagnosis have been observed among these women (9), and delays in treatment initiation are associated with poorer outcomes for mother and fetus (10). A possible intervention to avoid these adverse outcomes will be to promote active case finding for TB in pregnant women. This will, however, be appropriate only if pregnancy itself increases the risk of TB. In the United Kingdom screening is recommended for HIV-positive mothers and those with recent exposure to active TB only (12). However, literature from elsewhere has promoted the screening of the wider pregnant population (13).
There is currently insufficient evidence on the epidemiology of TB in pregnancy globally, and no data on whether pregnancy increases the risk of TB. The U.K. General Practice Research Database (GPRD) provided a suitable cohort to investigate the epidemiology of TB in pregnancy in the United Kingdom and to determine whether TB is an independent risk factor for TB to inform future public health and screening policy. Some of the results of the study have been previously reported in the form of an abstract (14).
We estimated incidence rates for TB in pregnancy, 6 months postpartum, and outside of pregnancy; modeled adjusted incidence rate ratios in a retrospective cohort study; and performed a nested self-controlled case series (SCCS) analysis, which adjusts for all non–time-dependent confounders (see analysis and Figure 1).
The study used the GPRD, which contains records from 460 practices across the United Kingdom, forming a generalizable data set of 5.5% of the U.K. population (15). Data derived from GP practices are audited, tested, and checked for quality assurance by GPRD staff and independent testers (15).
For the cohort study, all women with pregnancies occurring between 1996 and 2008 were enrolled in the study, with their entire individual cohort time potentially ranging from December 1987 to December 2009. Women had to have at least one clinical code (Read or OXMIS codes in the U.K. General Practice coding systems) associated with pregnancy in their medical records. We included all stillbirths, terminations, and miscarriages. Women with insufficient information to determine either start or end date of the pregnancy were excluded from the study.
For the nested SCCS approach, pregnant women with at least one Read code for TB were selected from the cohort. For individuals enrolled in this analysis, TB could have occurred before, during, or after pregnancy.
Pregnancy was defined as any conception, regardless of the outcome. The start of the pregnancy was defined using the recorded last menstrual period or calculated from the expected delivery date. The pregnancy end date was defined as the date of the recorded pregnancy outcome. If the start of a subsequent pregnancy overlapped with the end of a previous one, the start date of the second pregnancy was considered the end of the first pregnancy.
TB disease was either culture-confirmed disease caused by any species from the Mycobacterium tuberculosis complex or, in the absence of culture confirmation, the presence of clinical and/or radiological signs and/or symptoms compatible with tuberculosis, and/or a decision by the clinician to treat the patient with a full course of antituberculosis treatment.
These definitions were operationalized by choosing the respective Read codes from the database (TB events). We used the recording date for the TB diagnosis as our best estimate for the onset of TB disease. We counted repeated TB codes as new episodes if they were entered more than 12 months after the initial event (6). This rule excludes new episodes in women, who are currently receiving TB treatment. Socioeconomic status was measured using the U.K. index of multiple deprivations, a composite indicator based on employment, wealth, and assets in a small area (16).
Individual-level exposure factors, such as bacillus Calmette-Guérin (BCG) status, socioeconomic status, region, and marital status, were available from each patient's records. Time-bound exposure factors, such as age and time period (before 2000, 2001–2005, 2006–2010) were calculated on the basis of the mid-year point of the patient's birth year.
Individuals entered the cohort either 90 days after their first practice registration, the up-to-standard date of the practice, or reaching the age of 13 years, whichever came last. Person-time and events were censored at the end of practice registration (including death), last collection date for the practice, transfer-out date from the practice, or reaching the age of 50 years, whichever came first.
The database was cleaned and cross-checked. Other information from each individual's record, such as birth outcome, birth year, other pregnancies, death date, and practice transfers, was used to validate pregnancy and TB dates. The GPRD helped in cross-checking death data against U.K. death records. Pregnancies with identical start or end dates were removed as duplicates.
The crude incidence rates of TB disease in the cohort overall, during pregnancy, in the immediate postpregnancy period, and outside of pregnancy were calculated as the number of TB events per person-time at risk. The countrywide estimate was also weighted for various population sizes across U.K. regions.
We compared the incidence rates (IRs) in pregnancy and postpartum with the rate outside of pregnancy, calculating IR ratios. A Poisson regression model was built, allowing the effect of these three strata to be adjusted for confounding factors. Each was added stepwise to the model and retained, if it changed the IR rates for the strata and a likelihood ratio test was significant (P < 0.05), indicating the model with confounder explains the data better than the one without. Variables for which effect modification is clinically plausible were tested for interaction, taking the magnitude of the observed effect and the width of confidence intervals into account to decide whether associations are clinically as well as statistically significant.
To determine whether pregnancy is an independent risk factor for TB we performed a nested SCCS. This method, which has been validated elsewhere (17, 18), includes only individuals who have an event and exposure (n = 177; Figure 1). It compares the incidence of the event (i.e., TB) during the exposure time with the respective incidence in a “control time” (i.e., comparing incidence in the same person during a risk period with a nonrisk period). The exposure times in this study are times during pregnancy and the 6 months after pregnancy, and control times are all other person-times before and after pregnancy (Figure 1). The method implicitly controls for all non–time-dependent confounders of TB, such as country of origin or ethnicity. Age and period were adjusted for by including these variables in the model. The model was fitted by conditional Poisson regression and has similar statistical power compared with cohort studies (17).
We performed a sensitivity analysis by excluding the 6 months before each pregnancy from the calculation of background risk as per protocol and in keeping with the literature (8). This is because there is a theoretical possibility that women with worse health (e.g., unrecognized TB) are less willing to conceive, hence artificially lowering the pregnancy rate during this period. All data analysis was performed with STATA 11.1 (StataCorp LP, College Station, TX).
For the incidence study we estimated that about 35,000 person-years or 50,000 pregnancies would be needed to detect the 3-fold higher TB incidence rate in pregnancy seen in some local studies (8, 9) with 80% power and 5% significance, assuming a U.K. background rate of 13 per 100,000.
For the SCCS we estimated that 25 exposure-time events and 150 control-time events would be sufficient to detect a risk ratio of 1.7, whereas 100 and 700 events would allow detecting a risk ratio of 1.33 with 80% power at a 5% significance level.
A total of 192,801 women with a total of 264,136 pregnancies (1–14 per woman) were included in the cohort study. Of the 516,589 women with pregnancies in the GPRD, about 271,625 (52.6%) were excluded, because their pregnancies were outside of the 12-year observation period and a further 52,163 (10.1%) were excluded because pregnancy dates could not be accurately determined from their records. Table 1 shows that the geographical and age distributions of included women are similar to those for all pregnancies ever reported to the GPRD.
|Included Women (Pregnancies 1996–2008)||All Women (Pregnancies 1987–2009)|
|Scotland, Wales, and Northern Ireland||22,828||11.8||74,387||14.4|
|Age at start of first pregnancy|
Average follow-up time was 9.1 years (7 d to 21.8 yr), giving a total of 1,745,834 person-years. Of this time, 171,765 years were spent during pregnancy and 114,866 years in the 180 days postpregnancy. The median prepregnancy follow-up time was similar for women with TB (2.57 yr; interquartile range, 0.7–5.19 yr) and women without TB (median, 2.28 yr; interquartile range, 0.74–5.2 yr; P = 0.4).
The mean age at pregnancy was 29.5 years (range, 13–50). The mean age for cohort entry was 25.6 years (range, 13–48.9) and for cohort exit it was 35.1 years (range, 13.7–50). The median length of pregnancy was 39.6 weeks (range, 2–45), and two peaks of gestational length were observed, a smaller peak at 10.6 weeks and a larger peak at 39.9 weeks. Most pregnancies resulted in a not further specified birth (79%), a miscarriage (9.1%), a caesarean section (6.2%), or an assisted (e.g., instrumental) delivery (5.2%). There were 405 records (0.15%) of neonatal deaths.
A total of 177 TB events were identified in the cohort, 22 of these occurring during pregnancy (8, 7, and 7 in the first, second, and third trimester, respectively) and 22 in the 180 days after pregnancy. The mean age at TB diagnosis was 30.1 years (range, 13.4–44). None of the women had more than one TB episode. Most TB events were recorded in London (31%), followed by the Midlands (25%), north England (20%), south England (16%), and Scotland, Wales, and Northern Ireland (9%).
Of the TB events, 61.6% were not further specified. Among the remainder, extrapulmonary TB (n = 43) was more frequent than pulmonary TB (n = 25). This seemed more pronounced in the TB events during or 180 days after pregnancy, when extrapulmonary TB was found in 13 of 44 cases compared with 30 of 133 cases outside of pregnancy. The most common sites of extrapulmonary TB were genitourinary TB (n = 11) and lymphatic TB (n = 8).
Only one patient with a TB event in the cohort died and was censored 7 years after her pulmonary TB was diagnosed (outside pregnancy); it is improbable that the death was related to the TB.
The overall crude incidence rate of TB diagnosis was 10.1 (95% confidence interval [CI], 8.7–11.8) per 100,000 person-years. Taking into account the various representations of the regions, the weighted crude incidence rate over the cohort time (1987–2009) for the United Kingdom was 10.5 per 100,000.
The crude incidence rate for TB during pregnancy was 12.8 per 100,000 (95% CI, 8–19.4). The TB rate during the 180-day postpartum period (19.2 per 100,000; 95% CI, 12–29) was higher than outside of pregnancy (9.1 per 100,000; 95% CI, 7.6–10.8; P = 0.001). TB events during pregnancy and the 180 days postpartum combined (15.4 per 100,000; 95% CI, 11.2–20.6) were significantly more common compared with the rate outside of pregnancy (crude incidence rate ratio, 1.68; 95% CI, 1.17–2.38; P = 0.02).
TB incidence rate ratios (IRRs) adjusted for age, socioeconomic status, region of residence, and BCG vaccination status show a significantly increased incidence of TB in the 180 days after pregnancy (IRR, 1.95; 95% CI, 1.24–3.07; P = 0.004), but not during pregnancy (IRR, 1.29; 95% CI, 0.82–2.03; Table 2). The model also shows increased TB rates in London and more deprived areas and decreased rates among those with a record of BCG immunization (Table 2). Ethnicity and country of birth are not included in the Poisson model (insufficient data); however, these individual level confounders are adjusted for in the SCCS model.
|TB Events (n)||Person-Years||IR||95% CI (IR)||IRR||95% CI (IRR)||P Value||LR Test|
|Outside of pregnancy*||133||1,459,203||9.11||7.63–10.8||1.00||Reference category|
|6 mo postpregnancy||22||114,866||19.15||12–29||1.95||1.24–3.07||0.004||0.007|
|Age, 20–29 yr*||80||630,464||12.69||10.06–15.79||1.00||Reference category|
|Age, up to 19 yr||11||162,586||6.77||3.38–12.11||0.61||0.32–1.14||0.1|
|Age, 30–39 yr||74||778,319||9.51||7.47–11.94||0.74||0.54–1.02||0.06|
|Age, 40–49 yr||12||174,465||6.88||3.55–12.01||0.56||0.30–1.04||0.07||0.07|
|South England*||29||469,422||6.18||4.14–8.87||1.00||Reference category|
|Scotland, Wales, and NI||16||218,213||7.33||4.19–11.91||0.96||0.51–1.78||0.9||<0.001|
|Deprivation quintile 1*||12||341,793||3.51||1.81–6.13||1.00||Reference category|
|Deprivation quintile 2||18||304,738||5.91||3.5–9.34||1.50||0.72–3.12||0.3|
|Deprivation quintile 3||35||317,481||11.02||7.68–15.33||2.61||1.35–5.05||0.004|
|Deprivation quintile 4||53||367,564||14.42||10.8–18.86||2.98||1.57–5.65||0.001|
|Deprivation quintile 5||59||414,258||14.24||10.84–18.37||3.94||2.08–7.46||<0.001||<0.001|
The final SCCS model confirmed the results of the Poisson analysis (Table 3). Adjusting for all non–time-bound confounders, the time period of observation, and patients’ age, the incidence rate ratio of TB during pregnancy (IRR, 1.03; 95% CI, 0.64–1.65) is not significantly increased compared with the risk outside of pregnancy. However, the TB risk is significantly increased in the 6-month period after pregnancy (IRR, 1.61; 95% CI, 1.01–2.58; P = 0.04).
|TB Events||Person-Years||IRR||95% CI||P Value|
|Outside of pregnancy*||133||1,448||1||Reference category|
|6 mo postpregnancy||22||113||1.62||1.01–2.58||0.04|
|Age, 20–29 yr*||80||681||1.00||Reference category|
|Age, up to 19 yr||11||125||0.81||0.32–2.06||0.67|
|Age, 30–39 yr||74||742||0.79||0.44–1.42||0.43|
|Age, 40–49 yr||12||179||0.68||0.23–2.04||0.49|
|Before year 2000*||53||555||1.00||Reference category|
The postpartum period was explored in greater detail, using the same methodology. We found an upward trend immediately postpregnancy, peaking at 90 days postpartum (IRR, 1.74; 95% CI, 0.95–3.19) and a gradual decrease to an IRR of 1.53 (95% CI, 0.79–2.96) and 1.19 (95% CI, 0.55–2.58) at 180 days and 270 days postpartum, respectively (Figure 2).
Compared with the background risk, we did not find a significantly decreased TB risk in the 6 months before pregnancy (IRR, 0.53; 95% CI, 0.24–1.15; P = 0.11), and exclusion of this time period to calculate background risk had a minor effect on model estimates.
This is first primary care–based cohort study that quantifies the risk of TB during pregnancy and postpartum. We found a significantly increased risk in the 6 months after pregnancy, but not during pregnancy. Considering diagnostic, administrative, and immunological delays, TB risk during pregnancy is almost certainly also increased. The risk remained significantly elevated when adjusted for all individual-level risk factors and known time-dependent risk factors, as demonstrated in the SCCS.
This study benefits from using a large, representative, and well-maintained primary care database (15), but remains an observational study based on data recorded for clinical purposes. Our overall TB incidence rate was slightly lower than current U.K. TB rates (4) because it represents an average over the cohort time (1987–2009). Demographically, our cohort is similar to the entire cohort of pregnancies ever reported to the GPRD. Adjustment for confounding in the Poisson model was limited to well-recorded variables in the GPRD; however, these subject characteristics were adjusted for by design in the nested SCCS.
After cleaning and cross-validation, systematic misclassification is unlikely for exposure or outcome variables. Underascertainment of pregnancies is possible, because diagnosis and recording of early miscarriages can depend on health-seeking behavior. We may have underestimated TB risk during pregnancy, because TB is associated with adverse birth outcomes (10, 11, 19) and probably miscarriage. The effect of this is likely minimal and primary care data are probably less vulnerable to underrecording miscarriages compared with other data sources (20), and our descriptive analysis demonstrated that we captured many miscarriages. We limited analysis to women with pregnancies between 1996 and 2008 and those for whom start and end dates of the pregnancy could be ascertained. This could have led to underestimating TB episodes during pregnancy, especially in patients born abroad/in high-prevalence countries, because of late pregnancy presentations and poorer recording of miscarriages. However, recording of TB and pregnancies is done independently in U.K. primary care. Our sampling method is therefore unlikely to have introduced significant bias (Table 1).
The debate about an association between TB and pregnancy is unresolved (21), although early studies failed to demonstrate a clear association (22). Internationally there are no recent incidence estimates for non–HIV-infected pregnant populations. A previous United Kingdom–wide study of TB in pregnancy, based on an obstetric card return scheme (20) (4.2 per 100,000), may have underestimated incidence (23), because additional case finding was limited and it was restricted to obstetricians, who do not normally see miscarriages. Comparatively (8, 9) low extrapulmonary TB rates in this study (20) suggested an underascertainment of clinically less obvious cases. Our incidence estimates were lower than previous estimates from high-risk areas within London (8, 9), explainable by the different setting in our study.
Complex maternal immune system changes during pregnancy prevent allograft rejection of the fetus. One mechanism is the partial inhibition of the cellular immune system via cytokines (e.g., IL-10, transforming growth factor [TGF]-β) (24, 25) through an increase in specific T-regulatory cell populations. The susceptibility to specific infections such as influenza can increase during pregnancy (26). Mycobacteria cause a predominantly cellular immune response and studies have found an up-regulation of specific T cell–inhibiting cytokines (IL-10, TGF-β) during TB reactivation (27). The physiological down-regulation of the cellular immune system is one possible explanation for the higher incidence rates of TB during and immediately after pregnancy.
The significantly increased risk in the 6-month postpartum period may reflect a delay between pregnancy and risk increase for TB. Likely explanations fall into three categories—administrative, immunological, and medical. Because we took the first recorded TB date on the GPRD as the event date, an administrative delay between diagnosis and recording of TB disease is possible. It is also likely that immunological changes gradually increase TB susceptibility during pregnancy, as the expression of T-regulato cells is partially estrogen dependent (25) and increases throughout the pregnancy (25, 28). After delivery, these changes gradually normalize again.
Diagnostic delays have been described elsewhere (8, 9), despite the high level of access to health care enjoyed by pregnant women. Late presentations (8, 9); an ambiguity of symptoms, frequently mimicking physiological pregnancy changes (29); and a conservative approach to investigations (e.g., X-rays) have been blamed for these delays (30). Delays in treatment initiation are associated with poorer outcomes for mother and fetus (10, 19) and can be more pronounced in women of minority ethnic background or those who recently arrived from high-prevalence areas (31). These delays present a limitation to this study, and the combination of these delays is a possible reason for the failure to demonstrate a significant TB risk during pregnancy, and the anticipation of these prompted us to propose an analysis of the 6 months postpregnancy in our protocol.
In conclusion, we found that there was an increased incidence of TB diagnosis in mothers postpartum, which probably reflects an increase in TB incidence during pregnancy, given the diagnostic, immunological, and administrative delays described previously. About 25% of all events occurred during pregnancy and postpartum (16% of total person-time). To avoid adverse outcomes for mother and child it is vital to recognize and treat TB in pregnancy early (9–11, 19), making diagnostic delays unacceptable. This suggests that further research is needed on the cost-effectiveness of active case finding for TB in pregnant women. Research from Zimbabwe (32) found that untargeted active case finding in a high TB incidence area may be an effective tool for increasing case detection, and the role of simple and inexpensive interventions such as sputum smears during antenatal care should be evaluated in high TB incidence countries. Replication of our observations in cohort studies from TB high-burden countries may provide support for such an approach. In low-incidence countries clinicians should have an elevated index of suspicion for TB during the pregnancy and postpartum periods, aiming to exclude this diagnosis particularly in those from high-incidence countries. Our results may prompt rethinking of current TB prevention strategies and increase clinical awareness of TB during pregnancy and the postpartum period.
The authors thankfully acknowledge the funding and support of the Medical Research Council. In addition, the authors express their profound gratitude to Ms. Tarita Murray-Thomas and the staff at the GPRD and MHRA for their data help and support.
|1.||Sachs JD. Investing in development: a practical plan to achieve the Millennium Development Goals [Internet], 2010. Available from: http://www.unmillenniumproject.org/documents/MainReportComplete-lowres.pdf|
|2.||World Health Organisation. Global tuberculosis control [Internet], 2010. Available from: http://whqlibdoc.who.int/publications/2010/9789241564069_eng.pdf|
|3.||Ormerod P. Tuberculosis in pregnancy and the puerperium. Thorax 2001;56:494–499.|
|4.||Health Protection Agency. Tuberculosis in the UK: annual report on tuberculosis surveillance in the UK 2010. London: Health Protection Agency; 2010.|
|5.||Gilbert RL, Antoine D, French CE, Abubakar I, Watson JM, Jones JA. The impact of immigration on tuberculosis rates in the UK compared with other European countries. Int J Tuberc Lung Dis 2009;13:645–651.|
|6.||Crofts JP, Andrews NJ, Barker RD, Delpech V, Abubakar I. Risk factors for recurrent tuberculosis in England and Wales, 1998–2005. Thorax 2010;65:310–314.|
|7.||World Health Organisation. World health statistics 2010. Geneva, Switzerland: World Health Organization; 2010.|
|8.||Kothari A, Mahadevan N, Girling J. Tuberculosis and pregnancy: results of a study in a high prevalence area in London. Eur J Obstet Gynecol Reprod Biol 2006;126:48–55.|
|9.||Llewelyn M, Cropley I, Wilkinson RJ, Davidson RN. Tuberculosis diagnosed during pregnancy: a prospective study from London. Thorax 2000;55:129–132.|
|10.||Jana N, Vasishta K, Jindal SK, Khunnu B, Ghosh K. Perinatal outcome in pregnancies complicated by pulmonary tuberculosis. Int J Gynaecol Obstet 1994;44:119–124.|
|11.||Jana N, Vasishta K, Saha SC, Ghosh K. Obstetrical outcomes among women with extrapulmonary tuberculosis. N Engl J Med 1999;341:645–649.|
|12.||Health Protection Agency. Pregnancy and tuberculosis: guidance for clinicians [Internet]. Available from: http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/1194947408879|
|13.||Navarro-Alvarez S, Bobadilla del Valle M, Kato-Maeda M. Screening for tuberculosis in pregnancy. Int J Gynaecol Obstet 2002;79:251–252.|
|14.||Zenner D, Kruijshaar ME, Andrews N, Abubakar I. Pregnancy: a risk for developing tuberculosis? A national cohort and self-controlled case series study using UK primary care data [abstract]. Amsterdam: European Respiratory Society; 2011.|
|15.||Medicines and Healthcare Products Regulatory Agency and General Practice Research Database. Data Quality Assessment in GPRD. 2007.|
|16.||Department for Communities and Local Government. The English indices of deprivation [Internet], 2010. Available from: http://www.communities.gov.uk/documents/statistics/pdf/1870718.pdf|
|17.||Farrington CP, Nash J, Miller E. Case series analysis of adverse reactions to vaccines: a comparative evaluation. Am J Epidemiol 1996;143:1165–1173.|
|18.||Farrington P, Pugh S, Colville A, Flower A, Nash J, Morgan-Capner P, Rush M, Miller E. A new method for active surveillance of adverse events from diphtheria/tetanus/pertussis and measles/mumps/rubella vaccines. Lancet 1995;345:567–569.|
|19.||Lin HC, Lin HC, Chen SF. Increased risk of low birthweight and small for gestational age infants among women with tuberculosis. BJOG 2010;117:585–590.|
|20.||Knight M, Kurinczuk JJ, Nelson-Piercy C, Spark P, Brocklehurst P. Tuberculosis in pregnancy in the UK. BJOG 2009;116:584–588.|
|21.||Gupta A, Nayak U, Ram M, Bhosale R, Patil S, Basavraj A, Kakrani A, Philip S, Desai D, Sastry J, et al.. Postpartum tuberculosis incidence and mortality among HIV-infected women and their infants in Pune, India, 2002–2005. Clin Infect Dis 2007;45:241–249.|
|22.||Espinal MA, Reingold AL, Lavandera M. Effect of pregnancy on the risk of developing active tuberculosis. J Infect Dis 1996;173:488–491.|
|23.||Jana N, Arora N, Barik S. Tuberculosis in pregnancy in the UK. BJOG 2009;116:1542–1543.|
|24.||Saito S, Sasaki Y, Sakai M. CD4+CD25high regulatory T cells in human pregnancy. J Reprod Immunol 2005;65:111–120.|
|25.||Lin XG, Zhou Q, Wang L, Gao Y, Zhang WN, Luo ZL, Chen BC, Chen ZH, Chang S. Pregnancy estrogen drives the changes of T-lymphocyte subsets and cytokines and prolongs the survival of H-Y skin graft in murine model. Chin Med J (Engl) 2010;123:2593–2599.|
|26.||Siston AM, Rasmussen SA, Honein MA, Fry AM, Seib K, Callaghan WM, Louie J, Doyle TJ, Crockett M, Lynfield R, et al.. Pandemic 2009 influenza A (H1N1) virus illness among pregnant women in the United States. JAMA 2010;303:1517–1525.|
|27.||Bonecini-Almeida MG, Ho JL, Boechat N, Huard RC, Chitale S, Doo H, Geng J, Rego L, Lazzarini LC, Kritski AL, et al.. Down-modulation of lung immune responses by interleukin-10 and transforming growth factor β (TGF-β) and analysis of TGF-β receptors I and II in active tuberculosis. Infect Immun 2004;72:2628–2634.|
|28.||Wegienka G, Havstad S, Bobbitt KR, Woodcroft KJ, Zoratti EM, Ownby DR, Johnson CC. Within-woman change in regulatory T cells from pregnancy to the postpartum period. J Reprod Immunol 2011;88:58–65.|
|29.||Moore AR, Rogers FM, Dietrick D, Smith S. Extrapulmonary tuberculosis in pregnancy masquerading as a degenerating leiomyoma. Obstet Gynecol 2008;111:550–552.|
|30.||Centers for Disease Control and Prevention. Tuberculosis among pregnant women: New York City, 1985–1992. MMWR Morb Mortal Wkly Rep 1993;42:605, 611–605, 612.|
|31.||French CE, Kruijshaar ME, Jones JA, Abubakar I. The influence of socio-economic deprivation on tuberculosis treatment delays in England, 2000–2005. Epidemiol Infect 2009;137:591–596.|
|32.||Corbett EL, Bandason T, Duong T, Dauya E, Makamure B, Churchyard GJ, Williams BG, Munyati SS, Butterworth AE, Mason PR, et al.. Comparison of two active case-finding strategies for community-based diagnosis of symptomatic smear-positive tuberculosis and control of infectious tuberculosis in Harare, Zimbabwe (DETECTB): a cluster-randomised trial. Lancet 2010;376:1244–1253.|
Supported by a grant from the Medical Research Council (09_084R, 12/11/2009) to access the GPRD. Internal HPA funding was used to cover salaries and on-costs.
Author Contributions: D.Z. undertook the analysis and wrote the paper with contributions from M.E.K., N.R., and I.A. I.A. conceived the idea and N.R. supervised the statistical analysis. All authors have approved the final manuscript. I.A. is guarantor.
Originally Published in Press as DOI: 10.1164/rccm.201106-1083OC on December 8, 2011