There is no universally accepted methodology for assessing DFM. Every method has its limitations and a "gold standard" is difficult to define. Maternal perception of FM arises first and foremost as a result of pressure against body-wall structures, and thus the mother's perception reflects gross FM or limb movements [10,11]. The proportion of movements perceived by the mother and documented during ultrasound monitoring at the same time ranges from 37% to 88% [12]. A common factor in these studies is that the mother is lying down and focusing on fetal activity. This is the only situation in which we know that maternal perception and objective measures of FM are strongly correlated with objective measures of fetal activity. Outside such a setting, both the actual frequency of movements as well as the mother's ability to perceive them are influenced by factors such as maternal position [13], activity and exercise [14], anxiety [15], stress [16], blood sugar [17], smoking [18], placenta localization [10], and obesity [19]. Parity has not been found to affect maternal perception of FM in the third trimester [10], but multiparous might be able to perceive FM earlier in pregnancy than primiparous [20]. There are significant diurnal variations in normal fetal activity, which changes gradually with gestation [10,20].

Among the attempts to define DFM, a variety of methods of FM counting with different alarm limits have been published [1,6,7]. Among these, the rule of "ten movements within 2 hours" [21]. This is the only definition of DFM based on focused maternal counting which has been both developed and tested as a screening tool in a total population, and currently the definition of DFM recommended by the American Academy of Pediatrics and the American College of Obstetricians and Gynecologists if maternal movement counting is performed [22]. Other definitions of DFM have mostly been based on counting through both rest and activity and have little evidence in support of their association with actual fetal activity. The most important clinical understanding of DFM is still the mother's own perception of a decrease [1,23-26].

There are no universally accepted guidelines for the management of DFM [7,12]. Although several studies have presented guidelines including non-stress test (NST), ultrasound examination and Doppler [2,3,5,7,22,27], most of these recommendations are based on limited evidence, as we have reviewed elsewhere [7,12].

The number of women presenting with DFM remained unchanged during intervention at 6.3% versus 6.6% (OR 1.05; 95% CI 0.98–1.12, p = 0.19), respectively. The rate of unplanned repeat visits for DFM was consistently very low, but increased from 0.3% to 0.5%, p = 0.002.

The stillbirth rates among women with DFM were reduced by almost 50% (OR 0.51; 95% CI 0.32–0.81, p = 0.004)) from 4.2% (n = 50) to 2.4% (n = 73) during the intervention. Stillbirth rates among women in the entire cohort were reduced by one third from 3.0/1000 to 2.0/1000 (OR 0.67; 95% CI 0.48–0.93, p = 0.02). Independent data from the Medical Birth Registry in Norway, confirmed that the stillbirth rate in our total cohort of births was comparable to the rest of Norway in the baseline observation (OR 1.06; 95% CI 0.70–1.65, p = 0.73), and significantly lower during the intervention period (OR 0.64; 95% CI 0.47–0.87, p = 0.005). The intervention was followed prospectively with statistical process control charts which indicated a significant change in mortality after 7 months of intervention (arrow in figure ​figure2),2), and no month during the intervention with a mortality exceeding the pre-intervention mean (figure ​(figure2).2). There was no increase in secondary outcomes such as preterm births, FGR, severe neonatal depression or transfers to neonatal care among women with DFM during the intervention period (table ​(table11).

Among those with DFM, fewer women with a perceived absence of FM waited more than 24 hours, or a perceived decrease for more than 48 hours, before contacting health-care professionals during the intervention. There were no changes over time in the population in potential confounding factors as maternal age, BMI, smoking habits, parity or ethnicity (table ​(table22).

With a large prospective population-based cohort, a low "loss to follow-up" rate, a design with low risk of recruitment bias by outcome, ability to correct for anticipated confounders, large effects on hard outcomes, and confirmation of effects from independent data sources, the assessment of our intervention appear robust. Our quality assessment was conducted as a multi-intervention bundle that aimed to improve the in-hospital management of DFM, including clinical examination, the use of NST and ultrasound, recommended time-lines for health-care professionals, and excluding the use of Doppler. It included general information about fetal activity, recommendations for maternal care-seeking, several rules of thumb for recognizing DFM, and an FM chart as a supportive tool. It also included awareness among health-care professionals, since all obstetricians, general practitioners, community midwives, and others contributing to antenatal care in our population were informed in writing about the ongoing intervention. The exact effect size can only be estimated in randomized trials, which may be challenging and of moderate value unless each individual component of the bundle is tested in a separate trial [1,25]. Implementing only parts of the bundle as a response to the findings of our initial quality-assurance data was not an option in our high-resource setting with a highly educated population. It was considered unacceptable to inform women about DFM without securing professional management of DFM according to the consensus of best practice, and equally unacceptable to perform quality assurance of management of DFM without informing the women to the best of our knowledge about their important role in identifying and reporting DFM.

A much-debated issue is whether women should receive uniform information about FM, and whether this should include formal fetal movement counting (FMC) [25]. This is a method used by the mother to quantify FM, and the source of quantitative definitions of DFM, also called "alarm limits". Two main groups of counting methods exist, using either a "fixed time" or "fixed number" approach. The "Daily Movement Count" [33] reflects 12 hours of maternal FMC through an entire day (i.e., "fixed time"). This method was later modified to shorter and repeated periods of counting [1]. The "Count to ten" or "Cardiff" method uses the time it takes to perceive ten movements (i.e., "fixed number") [34]. The latter method is the most user-friendly, since a shorter time is needed to perform counting for normal pregnancies. This method has also been shown to have the highest compliance and acceptance rates [6,35,36]. While three controlled trials (one randomized) of FMC counting versus no counting has suggested benefit in preventing stillbirths [21,37-39], a large cluster multicentred cluster-randomized controlled trial reported by Grant, Valentin, Elbourne & Alexander in 1989 failed to demonstrate the same benefit using a "Kick Chart" for all pregnancies versus only for risk pregnancies [40]. This is the most referred-to and influential publication on maternal counting, and as such is often cited as evidence against FMC [1,28,41]. However, this trial had several of limitations [1,6]. Of greatest importance is the issue of contamination between the groups through the use of "within-hospital" clusters. The problem of contamination is compounded by the use of Kick Charts for control-group women on the basis of clinical discretion as a part of the trial design. While no difference was shown in the stillbirth rate across the study groups, the overall late-gestation stillbirth rate fell during the study period from 4/1000 to 2.8/1000 [40].

The lowered overall stillbirth rates seen in the observational cohorts and during the cluster-randomized trial might, however, be attributable equally to increased awareness and vigilance, as to the actual FMC methods and alarm limits. Indeed, the cluster-randomized trial used extreme limits (ten movements in 10 hours for two days or no movements for one full day) and based their "count to ten" method on the mother's perception through the day, and not on focused counting while lying down. Thus, the women needed 162 minutes to count ten movements versus the average of 20 minutes reported in focused counting [20,21,42]. Despite the extreme nature of such limits, they are still widely used [43]. There is no evidence that formal FMC with their fixed alarm limits are superior to maternal common sense, no evidence to support the introduction of such counting in any total population, and no rationale to perform trials using the existing alarm limits of FMC [25]. Better tools to identify the pregnancy at risk by assessing FM patterns are needed, and they will have to be individually adjusted to identify change, not fixed levels, to reflect what pregnant women are actually reporting. However, the routine of daily FMC in the third trimester could provide additional vigilance in the individual pregnancy, and help the expectant mother to identify significant changes. Our information highlighted the importance of the woman's subjective perception of a significant and sustained reduction in FM as the primary indicator of DFM, and a cause to seek professional help. We suggested daily FMC only as a tool to aid monitor FM, and guided the woman with "ten FM within 2 hours" as a secondary rule of thumb in situations where she felt in doubt.

The goal of antepartum fetal surveillance is to exclude imminent fetal jeopardy, identify risk pregnancies and aid in the prevention of adverse outcomes [27]. Controlled trials of management of DFM are lacking [7,12]. While the behaviour of health-care professionals related to the time of referral or examination remained unchanged during the intervention, the use of ultrasound changed. This was in accordance with the consensus-based guidelines of our study [12] indicating that NST and ultrasound examination were the most useful tools for fetal surveillance in DFM, and consistent with the evidence for antepartum testing in other risk pregnancies [12,44-46]

A weakness of the assessment of the intervention is that there are no codes for visits due to consultations for DFM in the electronic medical files of the Norwegian hospital system. Thus, no validation of the completeness of registrations of cases of DFM was possible with the anonymous of files used. Bias may have been introduced through the health professionals' inclusion of cases either by registration fatigue over time or increased enthusiasm by the general awareness caused by the intervention. This would, however, not affect the results on stillbirth rates in the total population, and not the outcomes among cases with DFM. Only a systematically skewed registration towards more or less severe cases of DFM would affect these results, and our design separating inclusion from outcome registration would counteract such effects. An additional weakness of the intervention is that we do not have the overall caesarean section and induction rate in the total population. However, it is unlikely that there would be any increase in the total population as the caesarean section rate following consultations for DFM remained unchanged and the induction rate was reduced. Clinical quality interventions in a population are based on the existing imperfections found by prior data collections of quality indicators, as we have demonstrated in our community. The results may thus not be directly transferable to other populations. Yet, reports from a variety of locations suggest that significant variability in the management of DFM and of information given to expecting women is a wide-spread quality issue in obstetric care [2,5,12,29].