Multiple reviews have linked environmental variables like proximity to public and private recreational facilities, presence of sidewalks, and neighborhood aesthetics to recreational physical activity and walking [4-7]. Walkable neighborhoods characterized by high residential density, proximity between homes and destinations, connected street networks, and pedestrian facilities such as sidewalks have been related to walking and cycling for transportation [4-6,8]. Neighborhood safety is an important aspect of the neighborhood social environment with potential influence on physical activity behaviours.

Safety is often measured as a simple undifferentiated construct, but for physical activity two aspects of safety are particularly relevant: traffic and crime. Studies on perceived crime and traffic safety as barriers to physical activity have produced inconsistent results [9-11]. For example, in a review [6], measures of personal safety were related to walking in the expected direction in six studies, but were related in the null or unexpected direction in ten other studies. Associations between aspects of safety and physical activity have also been reported to differ between subgroups, such as men and women [12-16], African American and White adults [17], older and younger adults [18] and overweight and obese men [19]. However, many of these studies have been conducted in Western high income countries, and further investigations to understand perceived and real safety barriers for different populations, especially from developing countries, are needed. Recent findings from the developing countries of Latin America suggest that relationships between environmental attributes and physical activity in the developing countries may be complex and may be different from those in high- income developed countries [20-22]. For examples, Parra et al [20] in Curitiba, Brazil found perception of personal safety from crime when walking and bicycling, but not safety from traffic to be positively associated with physical activity in adults, however, Gomez et al [22] in Bogota, Colombia found safety from traffic to be a positive correlate of more physical activity in older adults. In another study conducted in three regions of Brazil, no significant associations were found between physical activity and traffic conditions, and safety related to walking and bicycling during the day and at night [21].

To date, only one study has been conducted on environment--physical activity relationships in Africa, and walking was found to be associated with perceived safety from crime in the expected (positive) direction [23]. However, crime safety was related only to walking and not overall physical activity, and traffic safety was related with physical activity in the unexpected (inverse) direction in the Africa study. The study was limited to university students with potential higher socioeconomic status and younger age than the general population, and the safety items utilized were not specifically tailored to the African context. Thus, the one study in Africa reflects similar discrepant findings seen in previous studies. Since perception of safety between residents of different continents is likely to vary, the use of safety items from environmental surveys constructed in Western developed countries without local adaptation is a major concern, especially in Africa where the sociocultural and physical environments are distinct.

Perception of the neighborhood as safe from crime during the night emerged as the strongest correlate of health related physical activity. This finding was observed for both objective MVPA and self-reported walking. Consistent with some previous studies [13-15,50], this finding suggests that an unsafe neighborhood from crime at night may have negative influence on residents' physical activity. Feeling unsafe may diminish confidence in the ability to be physically active [15] or willingness to go outdoors. This is important in low income countries of Africa where neighborhood safety may be particularly salient.

The findings for safety from crime during the day were inconsistently related to physical activity outcomes, increasing the odds of self-reported walking but decreasing the odds of objective MVPA. This pattern is similar to discrepant findings in this literature in general [9,11,20,21]. Perhaps perceived safety during the day is related to walking because much walking takes place in the neighborhood, including required walking for transportation. Perceived safety from crime during the day may be related differently to total physical activity, which also reflects occupational or leisure activities that may be done elsewhere and influenced by different factors. However, it is possible that other environmental and/or personal factors unique to the developing countries may help explain the discrepant findings on associations between safety from crime during the day and physical activity outcomes in this study. Different environmental attributes and personal factors have been associated with different physical activity outcomes among adults in the developing countries [21,51], suggesting that these relationships are complex and may differ from those in high-income countries [20]. Additional study will be needed to explain present results.

Associations in the expected direction were observed between perception of safety from traffic for walking and meeting recommendations for objective MVPA. This result suggests that heavy traffic may be a barrier to physical activity and provides preliminary evidence of the need to provide safe traffic environments to support physical activity in Africa. However, our finding may not help to resolve discrepant findings across studies because traffic safety for bicycling was not related to objective physical activity in the present study. The lack of findings for safety to bicycle is not surprising because bicycling rates have been unstudied and appear to be low among Nigerian adults compared to the developed countries. Previous studies were also unable to provide consensus on how perception of traffic was related to physical activity [5,9,23,52,53], but this variable has not been studied extensively. Perhaps improved questions about facilities that protect bicyclists and pedestrians from traffic, and specific questions about traffic speed, traffic volume, and safety of intersections would provide more informative results.

There were a few gender differences in associations of safety variables with physical activity. While safety from crime at night appeared to be a barrier for men to engage in MVPA and walking, it was not relevant for womens physical activity. Men in Africa may have to return home at night from their occupation and be active outdoors, while women may rarely go out at night because of the societal expectation on them to take care of domestic related activities. However, women in this study tended to walk more when they perceived their neighborhoods as safe from crime during the day. Perhaps women are particularly likely to walk in the neighborhood during the day, so perceived crime during the day could inhibit this relatively common behavior.

Striking differences between men and women were found in the mean ratings of safety. On all four safety items, men reported feeling significantly safer than women. The biggest difference in perceptions, 66 percentage points, was found for traffic safety for walking, and this variable was correlated with objective MVPA for women. Because traffic volume and traffic speed may be differently related to physical activity [54], it is possible that high speed of traffic is an important barrier to physical activity that is specific to women. This evidence is an empirical rationale to implement and evaluate efforts to reduce traffic speed and aggressive driving on women and men. However, no definite conclusions on this topic can be drawn yet, because the present study had a cross-sectional design and it is the first from Africa showing this effect. Still the result shows traffic control is a potential health promotion and physical activity intervention in Africa.

There were several limitations of this study. Probability sampling was used to recruit participants from diverse neighborhoods, but the relatively small sample from a single city in Nigeria may limit generalizability. Walkability was not adjusted for in this study because walkability was confounded with SES, which reflects the reality in Nigerian cities. It is important that future studies develop neighborhood walkability and/or activity- friendliness indexes unique to Nigeria based on empirical analysis. The cross-sectional design does not allow us to determine causal relationships. Prospective or quasi- experimental studies are needed to strengthen the evidence of causality in this field. Although the scoring of objective physical activity at moderate-to-vigorous levels was consistent with other studies, [32,55], different methods have been used [56], and there is not a consensus about how to compute achievement of physical activity guidelines. The present study was conducted during two distinct seasons in Nigeria with varying weather conditions that could impact physical activity behavior. Though the perceived safety variables were specific to the neighborhood, neither of the physical activity outcomes was neighborhood-specific. This mis-match introduced error, but makes the several significant findings more notable.

The findings from the gender- specific analyses should be interpreted with caution as most of the resulting estimates were not precise. Small sample size can adversely affects precision of estimates and compromise the power of a study [30]. However, the findings can be considered preliminary and indicate the need for larger and more definitive studies, especially on effect modification of gender between neighborhood safety and physical activity in the African population.

Strengths of the study included separate measures for crime safety and traffic safety, objectively measured total physical activity, and self-report walking, which is the most common physical activity. The sample was selected to represent a range of neighborhood physical and socioeconomic environments. Though there are some inconsistencies in present findings about the relation of perceived traffic and crime safety to physical activity among Nigerian adults, the preponderance of evidence indicates that both crime and traffic safety appear to be significant barriers to physical activity. Because women felt less safe on all items, it appears that women are more affected by safety concerns.