Attitudes, practices and perceptions of KMC in this study showed significant improvement from admission through discharge and follow up visits among mothers. KMC has been found to promote breastfeeding in several studies,[7, 15, 16] and in this study breastfeeding was continued and sustained among most mothers. Ninety four percent of mothers were exclusively breastfeeding on discharge and they continued throughout the follow up period. These findings are consistent with a study in a tertiary care hospital in Brazil  where 108 (88%) LBW infants were exclusively breastfeeding on discharge with 87% still breastfeeding at 1 month. Whereas giving water and other feeds are discouraged during breastfeeding, the few mothers who continued to do so at discharge subsequently discontinued the practice on follow-up visits and counselling.
As mothers reported for follow up visits breastfeeding counselling was reinforced and mothers were more likely to change their wrong attitudes and practices. KMC has very high exclusive breastfeeding rates and where exclusive breastfeeding is uncommon among LBW infants, KMC may bring about an increase in breastfeeding prevalence and duration, with consequent benefits for growth and survival [8, 15, 17].
Perhaps not very surprising given the recent start of KMC in Ghana, very few mothers knew about KMC at recruitment. This notwithstanding, majority of them practiced KMC after the nurse had explained it to them. Two relatively young primiparous women were however still not comfortable with handling their tiny infants and as such declined to practice KMC. The mothers practicing KMC were also willing to recommend it to other mothers. These findings are consistent with results from a study in rural Ghana  involving 635 women from six districts. In that study, most of the women easily understood the KMC concept when they saw a picture of another mother practicing KMC and were willing to try it if it was good for the baby. In our study over 90% of mothers attested to the fact that KMC had been beneficial to them.
KMC practice outside the home was not acceptable to a relatively large proportion of mothers at discharge. The usual practice of carrying newborns in Ghana is for mothers to wrap them against their backs and not on their chest. In this study, many mothers at recruitment felt KMC will not be acceptable in the community because of this difference. However, KMC practice remained relatively stable with each visit. Mothers reported becoming more comfortable with the practice of KMC outside their homes and the proportion of helpers and spouses supporting them with KMC increased significantly with each follow up visit. The increased uptake of KMC could be linked to the fact that, the mothers as they said themselves, explained KMC to anyone in the community who asked about it. The community may have accepted the reasons given by the mothers, thereby indirectly reinforcing the mothers' resolve to continue KMC practice. These results are consistent with a study by Ruiz-Pelaez et al in 2004, which revealed that KMC produced a parental sense of fulfilment and improved confidence of mothers and caregivers as they were empowered by KMC to care for their preterm or low birth weight babies . Cattaneo et al  in three different tertiary hospitals in Ethiopia, Indonesia and Mexico, said KMC at all three facilities was considered feasible and mothers expressed a clear preference for KMC. This confirmed that in-hospital KMC for low birth weight babies was feasible in different settings, and acceptable to mothers of different cultures.
The mean weight gain of 23.7 g per day during follow up in this study was comparable to the study by Cattaneo et al  (21.7 g per day) but higher than that of Lima et al  in Brazil (15 g per day). The weight gain was lower in the study by Lima et al probably because it included averaged weight gain while on admission, the period for which neonates in our study experienced relatively no weight gain.
A major strength of our study is the low dropout rate of less than 20% (including one death) compared to what was expected over the four weeks of follow up. However, in spite of these significant findings our study has notable limitations. First, the mothers and babies lost to follow up could have biased the results, especially if mothers who discontinued KMC at home may have decided not to attend the follow up visits. Secondly, the mothers and caregivers were interviewed by healthcare personnel in a hospital setting. This may have influenced their responses at follow up visits. Also generalisability of the study is limited by the extra effort in the form of phone calls made to follow up mothers following discharge from hospital, which may have helped to maintain KMC. Finally, though the mothers were encouraged to practice continuous KMC because of its proven additional benefit compared to the intermittent, many of them chose to do the latter or combine the two. This could have negatively affected the established benefit of KMC to their infants.