In the United States, a substantial literature, much of it based on successive rounds of the National Survey of Family Growth, documents the incidence and correlates of unintended pregnancy and childbearing.¹ For developing countries, the Demographic and Health Surveys have provided a major database.² The essential approach to determining the intention status of pregnancies is similar in the two sets of surveys, though the questions differ. Pregnancies (or births) are classified as intended if they were wanted at the time they occurred or earlier, and as unintended if they occurred sooner than wanted or were unwanted at any time. The consequences, for the child and the family, of unintended pregnancies that are carried to term have also attracted research interest both in the United States and in developing countries.³

Published research on unintended pregnancy in China is limited, though studies of contraceptive failure and induced abortion have appeared in the literature.⁴ This neglect stems from the nature of government policy on family-size limitation. Since the introduction of the one-child policy in 1979, most couples in urban areas have been prevented from having more than one child, and most rural couples have been pressured to have no more than two children. It is, therefore, perhaps not surprising that more interest, at least internationally, has focused on desired childbearing that has been deterred by the state than on unintended births or pregnancies.

In this article, we start to redress this imbalance by analyzing unintended pregnancies, using data from a prospective study of newly married couples in Shanghai. Because the vast majority of couples in Shanghai are allowed to have only one child, we examine separately unintended pregnancy before and after first birth. For the period before first birth, the research questions concern the proportion of conceptions that occur earlier than couples wish and the correlates of such mistiming. For the period following the birth of the first child, the proportion and characteristics of couples who have unintended pregnancies are of interest. Additionally, we wanted to ascertain whether unintended pregnancies after the first birth reflect either family-size aspirations or son preference. We hypothesized that couples who would have liked to have a second child are more likely than others to conceive after having a child and, though a preference for sons over daughters may be less pronounced in Shanghai than in many other parts of China,⁵ that couples with a daughter are less satisfied than those with a son and thus more likely to wish for a second child.

METHODS

For logistic and cost reasons, the cohort was drawn from only two of Shanghai's 12 districts (Luwan and Hongkou). Luwan is a slightly more affluent area than Hongkou, but in general, Shanghai has a more homogeneous population than most cities in North America or Europe. The contraceptive profile of residents in the two districts is similar to that in the entire city: In 1998, roughly seven in 10 married couples used the IUD, and two in 10 used condoms; hormonal methods accounted for about 5% of use, and reliance on sterilization was negligible.⁶

Newly married couples were identified at the districts' marriage license offices between August 1987 and August 1988. All couples with an intention to delay the first conception after marriage were enrolled (7,220 couples), as were a random sample of 13% of couples without such an intention (691 couples).* Only 39 couples refused to participate.

Three months after their wedding, consenting couples were jointly interviewed at home by a trained field interviewer using a structured questionnaire. Other trained field interviewers, blinded to the previous data, reinterviewed the couple 15 months after the wedding and again during 1994-1995. A total of 7,872 couples completed the first interview, 7,826 completed the second and 7,693 the third; thus, the attrition rate over the course of the study was only 2%. Information on couples' demographic characteristics, sexual behavior, pregnancy histories and outcomes, and contraceptive use was obtained at each interview. Only couples who had had a live birth by the time of the third visit (92% of the original cohort) were included in the analysis because of the aim of comparing unintended pregnancy before and after childbirth. It should be stressed that voluntary childlessness is almost nonexistent in China.⁷

Both husbands and wives were asked whether each pregnancy after marriage was intended. The precise question was: "Did you want to become pregnant at that time?" If either spouse reported that a pregnancy was mistimed, it was regarded as unintended. We present results in terms of proportions of pregnancies that were unintended. In addition, we calculated the rate of unintended pregnancy by dividing the number of unintended pregnancies by the total couple-years of exposure to the risk of pregnancy in specific periods.

We performed bivariate analyses to assess associations between outcome variables and selected factors, measured as binary or ordered categorical variables; we used the chi-square test for trend to determine whether ordered categorical variables were linearly associated with unintended pregnancy. To control for confounders, we performed binary logistic regressions to assess factors predicting that a pregnancy before first birth was unintended, that a couple had a first unintended pregnancy after childbirth, and that a couple had two or more unintended pregnancies after first birth.

SPSS 10.0 and STATA 7.0 were used for this analysis. Data were weighted according to the initial selection probabilities, and weights were normalized to retain equality between the total unweighted and weighted sample sizes.

RESULTS

Fertility Preferences

By the time of the first follow-up interview, three months following marriage, 461 couples had conceived. When nonpregnant wives were asked how long a delay they had wanted between marriage and conception, 57% said three months or less, 33% said 4-12 months and 11% said more than one year (Table 1, page 7). Very strong relationships are apparent between the desired timing of the first birth and wife's age at marriage, the education and occupation of couples, and the education of the wife's parents. For instance, 23% of wives who married before age 24 wished to postpone conception for more than one year, compared with 2% of wives who married at age 30 or older. Nineteen percent of couples in which either spouse had a college education wished to postpone for more than a year, compared with 4% of those in which neither partner had progressed beyond middle school.

At the same interview, couples were asked whether they would have wanted a second child in the absence of the strict birth control regulations. In 50% of cases, neither the husband nor the wife would have wanted a second child. In 27% of cases, one spouse would have wanted a two-child family, and in the remaining 23%, both would have wanted a second child. Like the desired interval between marriage and conception, the desired number of children was related to the wife's age at marriage, the couple's education and occupation, and the educational attainment of the wife's parents. For example, the proportion of couples in which at least one spouse would have wanted two children was 44% if neither spouse had more than a middle school education and 54% if one or both were college graduates.

Unintended Pregnancies Between Marriage and First Birth

A total of 7,846 pregnancies occurred between marriage and first birth; of these, 21% (occurring among 1,543 couples) were reported as unintended. Most unintended pregnancies (81%) were caused by contraceptive failure rather than nonuse. The main contraceptive methods used to postpone the first birth were all ones with relatively high failure rates: withdrawal and periodic abstinence (which together accounted for 52% of episodes of use), and the condom (which accounted for 37%).⁸ The majority of unintended pregnancies (79%) were carried to term; 13% were terminated by induced abortion and 8% ended in miscarriage or stillbirth.

The proportion of pregnancies that were unplanned was very low—5%—among women who wanted to delay pregnancy for three months or less after marriage (Table 2). Among women who wished to postpone childbearing, the proportion was much higher—41% overall (not shown), ranging from 38% for those who wanted an interval of 4-6 months between marriage and conception to 47% among those who wanted to postpone conception for more than a year. Other factors were also significantly related to the occurrence of unintended pregnancy before first birth: The younger the wife at marriage and the higher the level of schooling of the couple or of the wife's parents, the higher the proportion of pregnancies that were reported as unintended; the proportion also was higher if one spouse had a white-collar job than if both were blue-collar workers.

When these relationships were reassessed using logistic regression to control simultaneously for all factors listed in Table 2, the wife's preference with regard to postponement of the first birth remained a strong and statistically significant predictor: Compared with pregnancies among women who wished to postpone conception for three months or less, those among women who wanted to wait longer had 13-16 times the odds of being unintended (odds ratios, 12.7-16.2). Age at marriage also remained a significant predictor; the odds of being unintended were about twice as high for pregnancies among women who had married at ages 20-25 as for those among women who had been 30 or older. However, the associations with education and occupation became nonsignificant.

Unintended Pregnancies After First Birth

After the birth of the first child, 3,167 couples (43% of the cohort) reported a total of 3,684 pregnancies, of which 97% were unintended.^† Eighty-nine percent of these couples experienced one unintended pregnancy, and 11% (or 5% of the entire cohort) experienced two or more. Virtually all of the unintended pregnancies (98%) were aborted; 1.7% ended in miscarriages or stillbirths, and 0.2% in live births.

The proportion of couples experiencing at least one pregnancy, the first of which was unintended, after having had a child, declined steadily as the wife's age at first birth increased and as the length of postpartum sexual abstinence increased (Table 3). Better educated couples had a lower incidence of unintended pregnancy than less educated couples, and a similar difference was apparent by occupation. Unintended pregnancy was slightly but significantly more common among couples in which one or both spouses would have wanted a second child in the absence of the one-child policy than among couples in which neither would have; its incidence did not differ by the sex of the first child.

The majority of couples had experienced no unintended pregnancy before the first birth, but 17% reported that the conception leading to the first birth had been unintended, and small proportions had experienced an unintended pregnancy that terminated in induced abortion (3%) or in miscarriage or stillbirth (2%). Unintended pregnancy after first birth was less common among couples who had terminated an earlier unintended pregnancy (24%) or whose earlier unintended pregnancy had ended in miscarriage or stillbirth (31%) than it was among couples with no prior unintended pregnancy (43%) or those whose first live birth had been unintended at conception (47%).

When reassessed by logistic regression, most of these relationships remained statistically significant. However, after adjustment for the effects of all other factors listed in Table 3, neither couples' education nor their occupation was significantly associated with the likelihood of experiencing an unintended pregnancy, and the desire for a second child was only marginally so.

Incidence of Unintended Pregnancy

The incidence of unintended pregnancy declined markedly as the amount of time since first birth increased (Figure 1, page 10). The highest incidence (32.5 per 100 couple-years) was recorded for the interval between marriage and the first birth, and most of these pregnancies were carried to term. In the first year following the initial birth, the incidence was still high—18.9. It then fell steeply until five or more years after childbirth, when it stabilized at a rate of 3-4 per 100 couple-years.

The proximate reason for this trend concerns the prevalence and nature of contraceptive protection. As noted earlier, contraceptive use between marriage and first birth was dominated by less effective methods. This was also true in the year after first birth, when 34% of exposed time was protected by condoms, withdrawal and periodic abstinence (not shown), and 20% by the much more effective IUD. The proportion of time protected by IUD use rose to 64% during the second year after the first birth and exceeded 70% in subsequent years.

DISCUSSION

Our analysis offers unique insights into the dynamics of reproductive control in Shanghai. As in all major cities of China, a one-child policy has been strictly enforced in Shanghai since 1979. Only a tiny fraction of couples have more than one child, usually following the death of the first child.

Like their counterparts in most Asian societies, couples in Shanghai generally do not wish to postpone childbirth following marriage. More than half of wives wanted a pregnancy within three months; only one in 10 wanted a delay of more than one year. Nevertheless, one in five pregnancies occurring before first birth were reported as unintended, a proportion similar to that found in neighboring Indonesia.⁹ Among couples wishing to delay pregnancy for more than three months, control of the timing of first birth was poor: Four in 10 had an unintended pregnancy, mostly because of contraceptive failure. Understandably, most of these pregnancies were carried to term. Educated couples and those with white-collar jobs were more likely than others to experience an unintended pregnancy at this phase of life. However, once the desired length of postponement was controlled for, the relationship between education or occupation and unintended pregnancy was not statistically significant; thus, it is largely explained by couples' desire to postpone childbearing.

All couples in Shanghai are well aware of the one-child policy. It might therefore be expected that pregnancy would be exceedingly uncommon following a first birth, except in the few cases in which the child dies. To the contrary, the analysis revealed a total of 3,684 pregnancies, nearly all of them unintended. More than four in 10 couples reported at least one unintended pregnancy, the vast majority of which were terminated. These pregnancies were concentrated in the 12 months following the first birth and, to a lesser extent, the subsequent year. For this reason, the duration of postpartum sexual abstinence emerged as a strong predictor of unintended pregnancy. Younger wives were significantly more likely than their older counterparts to have an unintended pregnancy after first birth, but surprisingly, neither education nor occupation was a significant predictor.

The few couples in which the woman had an induced abortion prior to first birth were much less likely than others to report an unintended pregnancy after childbirth. This finding, together with the fact that only 5% of couples had two or more unintended pregnancies following childbirth, suggests that repeated failures of reproductive control, leading to repeated abortions, are uncommon in Shanghai.

Does the occurrence of a pregnancy following first birth in a city where the one-child policy is rigorously enforced reflect the wishes of couples to have more than one child? In half of couples, at least one partner would have liked to have a second child; thus, adults in Shanghai apparently have little hesitation in articulating differences between the state requirement and their personal wishes. The likelihood of having one unintended pregnancy after first birth, and the likelihood of having two or more such pregnancies, were higher among pronatalist couples, but the differentials were not large and the associations were not always strong. The adjusted odds ratios show that couples wanting a second child had 12-14% higher odds of conceiving than other couples. The association with the likelihood of having more than one unintended pregnancy was more pronounced, but this outcome is rare.

Overall, we infer that the widespread resort to abortion by married couples in Shanghai reflects primarily not conscious or semiconscious reproductive wishes, but rather imperfect fertility control. This conclusion is strengthened by the absence of any link between sex of the first child and subsequent pregnancy, and is further buttressed by the close relationship between IUD use and the rate of unintended pregnancy. The proximate cause of the pattern is the type of contraceptive couples choose. Before and immediately following first birth, most couples use less effective methods; with the passage of time, they switch to IUDs. By the third year following first birth, nearly three-quarters of potential exposure to risk of conception is protected by this highly effective method.

Unintended pregnancies are common among married couples in Shanghai, mainly because of nonuse of effective contraceptive methods in the early years of marriage, both before and immediately following the first birth. Policies to reduce unintended pregnancies, and abortions, should focus particularly on postpartum contraception. Small-scale qualitative research is needed as a precursor to major policy initiatives, to establish why couples choose not to use any method or resort to ineffective methods.