Men and women in communities in Bauchi readily identified a range of factors they considered causal and preventive for kunika and indicated their views about the strength of the links between these factors and the outcome. They identified the most influential causes of kunika as frequent sexual contact, family dynamics, and non-use of both modern and traditional contraceptive methods. In separate maps of factors protecting against kunika they mainly identified as most influential the mirror-images of these causes.
Women and men had some different beliefs. Women, but not men, recognised forced or coerced sex as an important cause of kunika. Nearly all the women’s maps, but none of the men’s maps, mentioned lack of male involvement in reproductive health as a cause of kunika. On the other hand, men’s maps, and those of health providers at LGA and State level, placed more emphasis on increasing awareness about kunika and family planning methods as a way of reducing kunika.
The causes of kunika identified by women and men in Bauchi communities differ from those reported to be associated with short birth interval in quantitative studies in low- and middle-income countries. In a recent systematic review  only two factors were consistently associated with short birth interval in quantitative studies: shorter duration of breastfeeding and female sex of the previous child. The factors most frequently examined in the reviewed studies included age of the mother, education of the mother, education of the father, and household socio-economic status. Some of the maps in our study mentioned socio-economic status as a factor, but not a strongly influential factor. Age of the mother and education of the mother did not feature at all in the maps, perhaps because in the Bauchi context young age of mothers and low education of women are the norm, and do not explain why some women have kunika and others do not.
All groups in Bauchi identified frequent sex, with factors leading to this, as a key cause of kunika. Some maps specifically identified early resumption of sexual intercourse during breastfeeding as a cause of kunika. Qualitative studies, in Nigeria and elsewhere, have explored the dynamics of sexual contact after the birth of a child and during the period of breastfeeding; some have pointed to abandonment of traditional practices and taboos during this period as a cause of short birth interval [37,38,39]. Some of the dynamics around sexual intercourse between couples might have changed with the cultural transition towards more modern lifestyles and consequent changes in conservative sexual behaviour and social taboo [37, 38]. A study in the south of Nigeria found that loss of traditional healthcare practices was associated with more western education . However, most studies examining maternal education and birth interval have found shorter birth interval associated with less maternal education . The mapping groups in Bauchi talked openly about the use of aphrodisiacs and pornography as causes of frequent sex and hence kunika. Use of herbal aphrodisiacs is reportedly common among women in northern Nigeria [41, 42]. We are not aware of any other study reporting that men and women link the use of aphrodisiacs to kunika.
Participants identified use of contraception as influential in reducing kunika, counteracting the influence of frequent sex. Most studies reporting an association between contraceptive use and birth interval in low- and middle-income countries have found a longer birth interval among women using contraception . One multi-country analysis of data from Demographic and Health Surveys found shorter birth interval associated with modern contraceptive use, possibly because women began using contraception after experiencing an unintended short birth interval . Notes from the sessions discussing the summary maps in Bauchi indicated that participants felt people would not be willing to reduce sexual contact but would be interested to use contraception to make such contacts less likely to lead to kunika. Focus group discussions with women in Nigeria suggested that they rated the risks of short birth interval as greater than the risks of modern temporary methods of contraception .
All the mapping groups identified family dynamics as an important cause of kunika and identified modification of these dynamics as crucial to tackling kunika. These dynamics help to explain why people may continue to practise kunika, even when they know it has adverse consequences . The category includes the interaction between spouses (including interactions between co-wives), motivations to have more children and the influence of extended family and neighbours. Decisions about child spacing and family size are complex, and not well-reflected by the narrow concept of “unmet need for contraception”. Simply urging women to use contraception, without considering conflicting family dynamics, is unlikely to succeed.
Other studies in Nigeria have reported family dynamics that favour a short birth interval, such as a desire to have a large family, the need for children for a workforce, the influence of in-laws, and uncertainty surrounding children’s survival [45, 46]. Children as a source of male pride and competition between co-wives may favour short birth interval in Kenya . We considered “desire for a child of a particular sex” as part of the category of family dynamics. This was often, although not always, a desire to have a male child when the preceding child was a girl. Previous studies in Ethiopia reported a shorter birth interval if the previous child was female [47, 48], and this was also reported in some ethnic groups in Nigeria .
Forced or coerced sex featured as an influential cause of kunika in maps created by women, but not those created by men. Notes from the mapping groups suggested that often the women began by mentioning “love” as a cause of kunika, but later discussed that women could not refuse sex with their husbands and faced violence if they tried to. Levels of violence against women remain high in Nigeria and are higher than the national average in Bauchi . Less than half of women in Bauchi can refuse to have sex with their spouses . A recent multicounty study based on data from Demographic and Health Surveys found that emotional, physical, or sexual intimate partner violence was associated with shorter interpregnancy intervals and more unintended pregnancies .
Although not identified as having the strongest influence on kunika, “lack of male involvement” featured as a cause of kunika in 11/12 of the women’s maps of causes of kunika; but it was not present in any of the men’s maps (Table 1). This category included men not being willing to use contraception or get advice about it, and not letting their wives get such advice. In many African countries men are traditionally the main decision makers about when to have sex, whether to use contraception [50, 51], and how many children to have , although in recent years joint decision about contraception and fertility preferences has become more common . On the other hand, men’s maps gave more prominence to increasing awareness about kunika and family planning as a way of reducing kunika. This may suggest that they are aware of their lack of knowledge about these issues and are willing to get more involved.
Building and analysing fuzzy cognitive maps was part of our participatory research  into kunika in Bauchi State. Our partnership with community members, service providers, and health planners began by hearing the views and knowledge of community members through focus groups  and the fuzzy cognitive mapping described here. It went on to sharing the knowledge between stakeholders, and to proposing and implementing solutions. Our work on kunika was stimulated by the Bauchi State health authorities who expressed a concern about lack of birth spacing and suggested an exploration of how prevention of kunika could be incorporated into the home visits program. They are now planning to expand the home visits, including the kunika module, across the State.
Future research using fuzzy cognitive mapping could examine further the use of operator-independent methods for weighting of identified causes, which our findings suggest is a promising approach. It would also be interesting to examine in more detail the intermediate causes and common pathways by which different factors impact the outcome of kunika.
Fuzzy cognitive mapping does not mean the causes identified are “true” causes of the outcome, though they are true causes in the understanding of those who built the maps. The concern of soft models, however, is to engage stakeholders in identifying how to tackle the issue of concern, in this case, kunika. For the participants, these maps offer a way to present and to reflect on what they know. For researchers, the maps are sources of new hypotheses and variables to make sense of the issue.
All the participants in the mapping groups were married with children and most of them had at least two children but having two or more children was not an eligibility requirement. The participants shared their knowledge about the causes of kunika based not only on their own experience but also on what they knew of the experience of family and other community members.
We used transitive closure to calculate the influence of causal categories on the outcome. The compound maps present complex networks of interactions and it is difficult to consider all the reported relationships. The output of transitive closure can seem complex, even to researchers, so it was important to create summary maps that community groups can readily review and use. Our summary maps focussed on the strongest categories, and this simplification risks losing information.
Fuzzy cognitive mapping collects a lot of information in a short time. In our case, creating one map for causes and one for protective factors increased the time needed with each group. We had a break for lunch between the two maps, but the participants may have been tired and less engaged when creating the second map. Given the similar results using Harrisian discourse analysis, future mapping sessions might leave out the step of participant weighting which, though informative, can take as long as the building of the map.
It was not feasible to return to the communities and engage the map authors in the thematic analysis. The researchers who defined the broader categories of causes and protective factors used the mapping session notes and knowledge of the local fieldworkers to clarify the meaning of concepts. They checked their categories with the local research team and group facilitators. The assumptions and implications of our categorization are explicit in Additional file 1. The community groups presented with the summary maps agreed these maps fairly represented their views.