4

Quality Factors for Infant Formula: Normal Physical Growth

The statement of task asked the committee to examine the state of the science on methodologies to assess normal physical growth. In a public open session with the study sponsor (the U.S. Food and Drug Administration [FDA]), it further specified that the committee should consider benefits and limitations of the growth monitoring study (GMS) design as described in 21 CFR § 106.96(b) (see Chapter 2, Box 2-1). This chapter describes the historical regulatory context for using GMS to assess normal physical growth, evidence on use of GMS designs identified from the scoping review, and challenges with the current GMS methodology.

HISTORICAL PERSPECTIVE ON GROWTH MONITORING STUDY REGULATIONS

FDA regulations describe quality factors for assessing infant formula (21 CFR § Part 106.96 and 106.121), including the normal physical growth of the infant. FDA regulations require that normal physical growth be assessed through GMS, with specified requirements for study design and reporting, but allow for an exemption from the GMS requirements when “an alternative method or study design based on sound scientific principles is available to show that the formula supports normal physical growth in infants when the formula is fed as the sole source of nutrition” (21 CFR § 106.121).

The Infant Formula Act of 1980 discussed the need for quality factors to ensure healthy growth of infants (P.L. 96-359, September 26, 1980);

however, specific quality factors were not detailed. FDA’s 1996 proposed rule to implement the law, Current Good Manufacturing Practice, Quality Control Procedures, Quality Factors, Notification Requirements, and Records and Reports, for the Production of Infant Formula, interpreted “healthy growth” broadly:

The final rule (79 FR 33058, June 10, 2014) defined “quality factors” as those “factors necessary to demonstrate the safety of the infant formula and the bioavailability of its nutrients, as prepared for market and when fed as the sole source of nutrition, to ensure the healthy growth of infants.” FDA established the quality factor of normal physical growth because this was generally recognized as an indicator of infant health. FDA proposed two quality factors needed to evidence normal physical growth: (1) evidence that the formula supports normal physical growth of infants; and (2) the biological quality of protein used in the formula. Considerations regarding the biological quality of the protein, and a timeline for the quality factors regulations, were discussed in Chapter 3.

American Academy of Pediatrics 1988 Report

At the request of FDA, in 1986, the American Academy of Pediatrics (AAP) convened its Committee on Nutrition (CON) to recommend the types of clinical studies to assess the nutritional safety and suitability for healthy term infants that should be performed before marketing infant formula. The CON report recommended types of clinical studies and outcomes and circumstances that warrant clinical studies. FDA largely relied on it to support its quality factor of normal physical growth. However, some aspects of FDA’s 1996 proposed regulation go beyond the recommendations of the report. For example, the report states that “data on length gain are not considered essential in clinical testing of infant formula” because it is “unlikely that a significant difference in length gain between an experimental and control group will be demonstrated in the absence of a significant difference in weight gain.” Furthermore, since two persons are needed to obtain quality length measurements, “mandating measurements of length would be likely to result in accumulation of data without adequate quality control” (AAP, 1988). Similarly, CON consid-

ered and did not recommend measuring head circumference because “there is no evidence to suggest that change in head circumference will be useful in detecting the small differences in growth rate needed for clinical testing of infant formulas” (AAP, 1988). CON suggested potential value in recording food intake to assess efficiency of food utilization, such as fat (mal)absorption. However, the report concluded that studies that recorded food intake are “labor intensive and are unlikely to be feasible with the number of subjects needed to detect a small but nutritionally significant difference in weight gain between an experimental group and a control group” (AAP, 1988).

1996 Proposed Rule

The origins of “normal physical growth” as a quality factor were explained in the 1996 proposed rule (61 FR 36154, July 9, 1996). The quality factor reflected¹

Further discussion of the definition of quality factors and a timeline of actions can be found in Chapter 3 under the Evaluation of the Protein Efficiency Ratio section.

___________________

¹ FDA’s structure of naming the quality factor followed how it is demonstrated differs from the regulations. The biological quality of protein is named as the quality factor; PER (protein efficiency ratio) or an alternative method is the method to demonstrate this. As detailed above, “overall nutritional quality of the formula” should be the quality factor, for which normal physical growth is the indicator. However, FDA made normal physical growth the quality factor and GMS the method of demonstrating this.

The criteria for GMS included in FDA’s final rule were more stringent and frequent than recommended by CON, which was “that weight gain be determined over an interval of 3 to 4 months, beginning no later than 1 month of age” (AAP, 1988). The proposed rule followed the recommendation to enroll infants in the first month; however, in the 2014 final rule, FDA modified the enrollment age to not more than 2 weeks. CON also recommended that “measurements of weight should be made within well-defined age intervals, and should include an initial weight (e.g., at 14 days ± days)², a final weight (e.g., at 120 ± 4 days), and a weight at an intermediate age (e.g., at 60 ± 4 days)” (AAP, 1988). Instead, FDA required obtaining anthropometric measurements at least six times (once at the beginning of the study, once a month for the following 4 months, and once at the end of the study). The proposed rule stated:

The agency has tentatively concluded that the requirement of four measurements taken 1 month apart will provide a sufficient number of measurements to permit evaluation of whether the formula meets the nutritional needs for physical growth of the infant throughout the study period. However, the agency requests comment, supported by data, on which measurements are needed to provide evidence that the formula meets the nutritional needs for physical growth of infants (61 FR 36154 at 36184).

FDA Actions and Response to 1996 Proposed Rule

In 2002, FDA convened the Infant Formula Subcommittee of the Food Advisory Committee to explore questions relevant to review of infant formula (67 FR 12571, March 19, 2002; 67 FR 63933, October 16, 2002; 68 FR 8299, February 20, 2003), which held a series of meetings to discuss the scientific issues involved in evaluating whether a new formula supports normal physical growth.³ FDA also formally reopened the comment period once, in 2003 (68 FR 38247, June 27, 2003); it requested comments about a need for quality factors in addition to protein quality and normal

___________________

² The report does not show a number for the allowable variation surrounding 14 days.

³ The advisory committee’s findings were not publicly available.

physical growth, when GMSs were needed, whether to use the U.S. Centers for Disease Control and Prevention (CDC) or Iowa growth charts as the reference in GMS, and enrollment age, given that its advisory committee had recommended not more than 14 days. FDA also determined that it would remove certain details of GMS requirements from regulation and instead provide guidance on those details.

Acting on public comments submitted in response to the 1996 proposed rule and in these meetings of its Food Advisory Committee, FDA revised some of the proposed regulations on GMS requirements. For example, commenters recommended the inclusion age be revised from less than 1 month to not more than 2 weeks “because this early period is the time of greatest nutrient requirement and greatest sensitivity to nutrient adequacy” (79 FR 7934). Another comment suggested that enrollment should start “by 14 days of age in order to ensure a 4-month observation period before other foods are introduced into the infant’s diet” (79 FR 7934).

FDA agreed with these comments for reasons of data quality and practicality. It was thought that collecting data at this earliest phase of infancy would reflect the greatest nutritional demands on formula, have the greatest sensitivity to detect deficiencies in normal physical growth (Fomon et al., 1999), and avoid the potential effects of complementary feeding at around 4 months. Additionally, aligning the age of enrollment to the timing of well-baby visits was thought to potentially enhance recruitment.

FDA rejected other comments on the proposed rule. For example, a comment questioned the value of length measurements, citing the CON report. FDA stated in the 2014 interim rule that “changes in length and head circumference data provide especially valuable information for interpretation of the weight change data in those situations in which weight change with a test formula is significantly different than the weight change attained with the control formula” (79 FR 7934). However, FDA did not explain how these data would be used.

In response to a comment challenging the need for data on head circumference, FDA wrote that the reference to CON was not relevant because it “applies to a situation in which no significant difference is observed in weight change.” FDA explained that:

FDA did not explain how head circumference data would aid assessment of physical growth if a difference in weight gain was already established. Comments also questioned the requirement for measurement at six times over the study. CON had recommended three time points. FDA responded that the “proposed measurement schedule would accurately capture the curvilinear nature of infant growth and would provide sufficient data to interpret differences in growth and in growth rates, if differences exist” (79 FR 7934) but offered no explanation of why other times of data collection were not sufficient and how the “sufficient” data could be used to interpret differences in growth.⁴

Responding to comments questioning the need to plot individual growth data on growth charts, FDA wrote that it “provides an early indication of a possible problem with formula composition because it allows the investigator to observe disturbances in the growth of individual subjects” (79 FR 7934). FDA also demonstrated increased flexibility in its final rule, such as by recognizing the usefulness of studies conducted outside the United States and allowing for confidence intervals to estimate needed sample size of studies. In the reopening of the comment period in 2003, FDA indicated its intention to issue guidance on GMS design, but none has been published as of November 2024.

2004 Institute of Medicine Report

The most recent independent, authoritative U.S. body to discuss assessment of infant growth was the Institute of Medicine (IOM), which issued an FDA-commissioned report in 2004 to guide the assessment of safety of ingredients intended for use in infant formula. Growth was considered a fundamental measurement, which was recommended to be used in combination with other assessments. The IOM report recommended the growth assessment “for the purpose of determining the safety of an ingredient new to infant formulas.” It “must therefore be based on a longitudinal study, with repeated measures at relatively frequent intervals during the period when growth is most rapid and during the time period when formula serves as the sole source of infant nutrition” (IOM, 2004). The IOM report also recommended data collection on infant weight velocity, length velocity, and head circumference. The IOM report acknowledged that it was difficult to define normal growth but indicated that measurements below the 5th percentile or above the 95th percentile were typically cause for concern (IOM, 2004).

___________________

⁴ FDA rejected some proposed changes to the regulations in part because public comments provided no supporting data.

The IOM report considered normal physical growth to be a minimal assessment, as neurodevelopmental and organ system deficiencies can occur even when physical growth is normal. Additionally, body composition was recommended to be assessed even while acknowledging that it was difficult to measure accurately (IOM, 2004). The IOM report also recommended that both a formula-fed control group and a human milk–fed reference group be used as comparators for the new formula and that formula intake should be measured.

The IOM report also examined the magnitude of difference in weight gain that could be considered a safety concern: “There is very little scientific evidence to establish a level of difference associated with long- or short-term health consequences” (IOM, 2004). The IOM committee considered the level of 3g/d that was identified by the CON report, concluding that “no specific evidence to support this level of difference was provided.” The IOM report cites Fomon’s description of failure to thrive as a two standard deviation loss in weight over a 2-month period and a weight for length value below the 5th percentile.

The IOM report also cited studies comparing growth of formula-fed and human milk–fed infants. Dewey (1998) found greater weight velocity over 4 months in formula-fed boys but no difference in girls. Kramer et al. (2002) reported that infants weaned within the first month of life had lower weight and length z-scores (around 0.2 standard deviations) than exclusively human milk–fed infants. Hediger et al. (2000) used data from the Third National Health and Nutrition Examination Survey (NHANES) and did not find a difference in weight status by feeding method in 4–7-month-old infants but did in 8–11-month-old infants. The IOM committee observed that the 3g/d difference was approximately the difference between the population 25th and 50th percentile and that the clinical or functional significance of such differences is not well established. The IOM committee also expressed concern about excess weight gain in infancy as a risk factor for later life overweight or obesity, although studies examining that relationship were inconsistent. Finally, the IOM report commented on the duration of any growth study, noting the sensitivity of growth in the first 120 days, but the identification of adverse effects does not need to be confined to 4 or 6 months of life. Rather, the report suggested that adverse effects should be monitored during the entire duration of infant formula feeding.

CURRENT REGULATORY REQUIREMENTS

The requirements to conduct and report a GMS are given in 21 CFR § 106.96 (a) and (b); details of reporting requirements are specified in 21 CFR § 106.121 (a), (b), (c), (d), & (e). 21 CFR § 106.96(b) specifies that:

Regarding the general requirement for GMS (21 CFR § 106.96(b)), the committee noted that the terms “adequate” and “well-controlled” are not further defined in the regulations or any accompanying publicly available FDA guidance document. While FDA provides guidance on good clinical practice from the International Council on Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) (FDA, 2023), the committee is uncertain if the reference in 21 CFR § 106.96 to “good clinical practice” implies adherence to a specific framework, such as the ICH guideline on Good Clinical Practice (ICH, 2023).

While it may be inferred that some of the specific requirements within 21 CFR § 106.96(b), such as for a comparison of a “test group” and a “concurrent control group,” are an indication of the FDA criteria that constitute adequacy of the study, FDA provides no further elaboration of conventional GMS design requirements. Examples of specific criteria that may relate to study adequacy include eligibility (beyond age at enrollment); randomized allocation of infants to the test or control group; concealment of allocation and blinding/masking of study participants and investigators; characteristics of an appropriate formula or feeding practice used in the control group; non-inferiority designs with specification of a non-inferiority margin for the primary outcome and accompanying power calculation; and governance standards, such as trial registration or monitoring by an independent body, such as a Data and Safety Monitoring Board. These elements are described in some detail by regulatory agencies in other jurisdictions, such as Health Canada’s “Growth and tolerance clinical trial protocol” (Health Canada, 2021) and the European Food Safety Authority’s (EFSA’s) “Scientific and technical guidance for the preparation and presentation of a dossier for evaluation of an infant and/or follow-on formula manufactured from protein hydrolysates” (EFSA, 2021). However, as noted, FDA has not released a similar guidance document that clarifies its intended or actual implementation of 21 CFR § 106.96(b).

In a submission for a new infant formula, manufacturers provide to FDA the findings of a GMS that compares it to a similar formula that has already been reviewed by FDA (webinar 4 November 17, 2022, in the 15th minute) and is available on the U.S. market. As a matter of practice, the GMS is typically a randomized controlled trial (RCT); however, the committee was unable to identify detailed information on how FDA applies specific requirements to the design features of such RCTs or considers

applications that lack a GMS (an exemption under 21 CFR § 106.96(c) (2)(i)) or have a GMS that is not an RCT (e.g., use of historical or non-concurrent controls). As 21 CFR § 106.96(c)(2)(i) allows “an alternative method or study design that is based on sound scientific principles,” manufacturers may request an exemption from these requirements based on “a detailed description of the alternative method or alternative study design, an explanation of why the method or study design is based on sound scientific principles, and data that demonstrate that the formula supports normal physical growth in infants when the formula is fed as the sole source of nutrition” (21 CFR § 106.121(c)). For example, FDA could exercise enforcement discretion under emergency circumstances, such as the 2022 infant formula shortage. In such a situation, a company may have completed studies outside the United States that showed that the “formula supports normal physical growth in infants.”

Design and Reporting of Growth Monitoring Studies—Requirements by 21 CFR § 106.96

A summary of insights into the extent to which the evidence supports each of the regulatory requirements outlined in 21 CFR §106.96 is provided below. The committee’s assessment is based on the scoping review of clinical studies in infants that include growth monitoring, information from workshop speakers, and review of guidance clinical trial protocols and guidance documents from national and international regulatory agencies in other countries.

1. Duration of observation to be not less than 15 weeks

The current FDA regulatory requirement is that trials conducted to provide evidence “that a new formula supports normal physical growth” be “no less than 15 weeks in duration” (21 CFR §106.96(b)(1)). This stipulation ensures that infants enrolled by 2 weeks of age would be about 4 months at study exit, when about one-third of infants have already been introduced to complementary foods (CFs) (Chiang et al., 2023). This requirement differs from Canadian guidelines, which call for 16 weeks, and EFSA’s recommendation of 3 months (EFSA, 2021; Health Canada, 2021). The 2004 IOM committee acknowledged major limitations of FDA’s 15-week minimum study duration (which it referred to as the “120-day growth study” under consideration by FDA at the time, as the infants would be 120 days of age after completing the 15-week study if enrolled at 15 days of age). The IOM committee noted that when infant formula is used as a human milk substitute, it is likely to remain the major source of macro- and micronutrients even after introduction

of CFs and that the duration is too short to observe later health effects of the infant formula.

The committee’s scoping review did not identify any studies that attempted to assess the duration of observation on outcome; studies ranged from 4 weeks to more than a year. About three-fourths of the studies had feeding durations of 15 weeks or longer, which was common and of similar frequency in both U.S. and non-U.S. studies (see Appendix E).

One of FDA’s concerns in establishing the duration of GMS at 15 weeks was to minimize the confounding effects of introducing CFs (79 FR 7934 at 8012). In the context of a randomized trial, this would likely be similar across all treatment groups and therefore not be expected to bias treatment effects; however, it is possible for the formula intervention to affect the timing of introduction or amount of CFs consumed. Also, infants would not be excluded from an intent-to-treat analysis on the basis of consuming CFs before study exit. While extended follow-up of participants in GMS could identify longer-term health effects, it may be impractical to implement a requirement for long-term follow-up for a trial with a primary outcome of weight gain focused on initial approval of a new formula. Considerations of longer-term outcomes not limited to growth or weight are addressed in further detail next.

2. Age at enrollment of infants no more than 2 weeks old at time of entry into the study

The only GMS participant eligibility criteria included in current FDA requirements are that enrolled infants be “no more than 2 weeks old at time of entry into the study” (21 CFR § 106.96 (b)(1)) and that infant formula is “fed as a sole source of nutrition” (21 CFR § 106.96 (b)). Considered together, the implication is that eligible infants must have entirely ceased consumption of human milk, or never consumed it, by 2 weeks of age. Acknowledging the extensive discussion of this controversial requirement in public and expert consultations over nearly 3 decades, the committee was concerned about two related problems. First, nearly 80 percent of U.S. infants receive some human milk at the age of 1 month (CDC, 2022), underscoring the challenges with identifying eligible infants for enrollment. The committee heard testimony during open sessions from industry professionals who described the challenges that sponsors face in meeting enrollment targets because the vast majority of 2-week-old infants are consuming some human milk (CDC, 2022). The second problem is that this requirement may undermine the representativeness of the GMS participant sample in relation to the broader population of infants who would consume the infant formula if approved and sold. Estimates from CDC suggest that at 2 weeks of age, about 60 percent of infants are

exclusively human milk fed and 40 percent are receiving some formula; of the latter group, about half are still receiving some human milk (CDC, 2022). Therefore, enrollment of only those infants for whom the formula is the sole source of nutrition suggests that GMS participants are unlikely to be representative of about half of the infants who may eventually consume it. The committee recognized the FDA concern that intake of human milk by infants would interfere with assessing the nutritional adequacy of the new formula as a sole source of nutrition, but this must be balanced against the concern that the results of the GMS would not be generalizable to the population to whom the formula would eventually be sold.

The committee did not identify evidence from published peer-reviewed research that further informed its deliberations on this issue. The committee could find no studies that analyzed age of enrollment as a variable. Few studies examined growth from <15 days of age to that from older initial ages (Wallingford et al., 2022). The committee considered an alternative approach whereby FDA could extend the maximum age limit of enrollment to 3 or 4 weeks but retain the requirement that all infants be entirely infant formula fed; however, as mixed formula and human milk feeding remains highly prevalent at these ages (CDC, 2022), this modification would not substantially address the problems of feasibility and representativeness. Furthermore, the later the age of enrollment, the greater exposure that infants in the test group would have already had to either human milk or an approved, rather than the new, infant formula. The former could accentuate differences later attributable to formula, and the latter could attenuate between-group differences and undermine the applicability of the GMS results to infants who may start consuming the new formula as early as the day of birth.

According to the most recent CDC estimates, about 20 percent of infants are solely fed formula from early life, and this percentage does not increase substantially from 2 to 4 or 6 weeks of age (CDC, 2022). Thus, about 80 percent of U.S. infants are exclusively or partially breastfeeding. This suggests there would be no meaningful increase in the recruitment pool by revising the inclusion criterion from exclusive formula feeding from “less than 2 weeks” to “by 1 month,” although willingness to participate may differ if enrollment is allowed at older ages. However, the committee recognized that the trade-off of requiring earlier enrollment is the risk of interfering with breastfeeding promotion, as many infants exposed to infant formula early in life may still be able to exclusively breastfeed thereafter with appropriate lactation support.

The current FDA regulations do not discuss GMS eligibility criteria (other than age at enrollment and being solely formula fed) or include a requirement to ensure that the study population is representative of the general U.S. population of infants who would be likely to consume the

formula if approved and marketed for sale. The committee considered the risk that without such guidance from the FDA, GMSs may lack population representativeness and, in particular, disproportionately exclude vulnerable or marginalized populations. Additional GMS design requirements that are specifically related to sub-items in 21 CFR § 106.96(b) and § 106.121(a) are explored next.

3. Collection and maintenance of data on formula intake and anthropometric measures of physical growth, including body weight, recumbent length, head circumference, average daily weight increment, and average daily recumbent length increment

The 21 CFR §106.96 (b)(2) requires that a manufacturer include in its GMS collection the maintenance of data on infant formula intake and anthropometric measures of physical growth, including body weight, recumbent length, head circumference, average daily weight increment, and average daily recumbent length increment. The committee did not consider the necessity of each measure separately (21 CFR § 106.96(b)(2)), other than to note that weight was the measure most relied on and that average daily recumbent length was added in the final rule in response to one comment that asserted that daily increments in weight and length were the most sensitive measures of infant growth. As noted in the discussion on statistical analyses, well-accepted methods are available that do not rely on these daily increments of gain in weight or length. The literature scan did not identify any methodologic studies that assessed the usefulness of individual required or calculated measurements in evaluating normal physical growth other than that weight gain increments over the study duration were commonly used to estimate sample size.

During most GMSs, infants are commonly brought to a clinical research facility where trained study personnel conduct anthropometric measurements according to standardized operating procedures (SOPs) and using standard devices/equipment (e.g., precise and calibrated weighing scales, length measuring boards). Training personnel and standardizing the methods for anthropometric data collection is essential. Globally, the method for each of the anthropometric measures and assessment of staff performance precision follow the guidelines for standardization and training available from the World Health Organization (WHO) (WHO and UNICEF, 2019). Alternative approaches include the location of data collection and the individual collecting the data. Each has its own limitations. In situations where travel to a research center or hospital is not practical (such as during the COVID-19 pandemic), alternative approaches to obtaining anthropometric measures have included parents/caregivers or health professionals collecting the data, such as in

a doctor’s office (Lasekan et al., 2022; Parschat et al., 2021), but detailed information was not included on the training and standardizations for conduct of the measures. In such community-based collection of primary outcome measures, extensive training of data collectors to follow the WHO guidance on conducting the measures, distribution and calibration of the measurement tools, and close oversight of precision of the measures are required. WHO and the United Nations International Children’s Emergency Fund (UNICEF) have stated the importance of anthropometric data quality assessments in the preprocessing of raw data collected in a GMS (WHO and UNICEF, 2019), which may include visualization and statistical description of data distributions, identification and reconciliation of outliers, and specific tests to examine for evidence of data manipulation (e.g., heaping due to digit preference) (WHO and UNICEF, 2019). FDA regulations do not mention data quality assessments, but FDA representatives informed the committee that such assessments are routinely undertaken by FDA analysts who have access to raw data submitted by manufacturers (NASEM, 2024a).

4. Timing of anthropometric measurements

The 21 CFR § 106.96(b)(3) requires:

The committee was not aware of how FDA has used the data for measurements between the initial and final measurements in its evaluation of normal physical growth. FDA offered the example of an accepted “alternate method or study design” as a study that had only five data collection time points, suggesting that using six time points of measurement is not essential. The scoping review conducted by the committee found that 81 of the 143 studies reported weight fewer than six times, whereas only 33 used six or more time points, suggesting that investigators determined that six time points were not necessary. Fewer studies from outside the United States than within reported three data collections in the first month of the study.

Another reason FDA proposed six measurement time points was that it needed

However, the agency did not require the use of statistical methods that assess curvilinear growth but only the comparison of group means at each measurement time. A methodologic comparison has not been performed to establish any advantage to the data collection regimen that FDA requires manufacturers to demonstrate normal physical growth. Similarly, FDA made the strong case that because of rapid growth in the first weeks of infancy, the measurements at 2 and 4 weeks of feeding were required because they had greater sensitivity to detect nutritional adequacy. Sensitivity at particular time points should be considered in the context of other aspects of study design; for example, a feeding duration of at least 15 weeks allows cumulative effects to be evaluated, offering a sensitivity to detect small directional changes over time.

The committee was unable to identify information on FDA’s review process to illustrate how data from intermediate collections were used; it was not aware of instances where FDA determined a formula to be nutritionally inferior (i.e., significant difference in attained weight or in weight velocity) based on any single time point before study completion. The committee did not attempt to systematically identify the occurrence of significant differences at intermediate measurement times. Anecdotally, it identified through the scoping review one report with significant differences in some anthropometric measure at an intermediate time, albeit not at the end of the study (Xia et al., 2021). It is not clear how FDA would interpret a transient difference in attained weight or weight velocity given no difference at the end of the study.

Limitations to the current paradigm include needing to collect data that may not be necessary to determine the nutritional adequacy of a formula to support normal physical growth, adding to study complexity, cost, and invasiveness. Another weakness is the probability of detecting a statistically significant difference by chance, because of the large number of statistical tests. This likelihood is increased when analyses are done on each sex independently. A weakness is the interpretation of transient differences, which may not reflect nutritional adequacy of the formula.

5. Comparison of anthropometric measurements between groups

The 21 CFR § 106.96(b)(4) requires the GMS to compare:

As acknowledged by the 2004 IOM committee, a feasible albeit indirect method of comparing formula-fed to human milk–fed infants is to standardize the anthropometric measurements using growth references/standards that have been developed using data from exclusively or predominantly human milk–fed infants (IOM, 2004). FDA, Health Canada, and EFSA, support the comparison of GMS data to a suitable national or international growth reference/standard; FDA identifies it specifically as the 2009 CDC charts (per 21 CFR § 106.96(b)(4)). The CDC charts are based on the WHO child growth standards for children 0–2 years of age (CDC, 2024a), which were developed from a multi-country longitudinal cohort of infants who met stringent dietary criteria, including “exclusive or predominant breastfeeding for at least 4 months, introduction of CF by the age of 6 months, and continued partial breastfeeding up to at least 12 months” (WHO, 2006). The WHO standards established an international normative standard of early-childhood growth and have served as an invaluable tool for anthropometric data analysis in epidemiological studies and clinical trials worldwide.

The 2004 IOM committee expressed its concern about the definition of “normal physical growth.” The WHO child growth standards were not released until after the IOM report, and no similar international standard was available at the time. Nonetheless, the standards cannot be assumed to perfectly reflect the distributions of weights and lengths that enrolled infants would have attained had they been exclusively or predominantly fed human milk in the particular populations and settings in which every GMS is conducted; and some divergence of the average trajectory of the formula-fed infant from a human milk–fed standard is expected (Dewey, 1998; Wallingford et al., 2022). However, the standards provide a practical method of assessing whether the deviations of the average growth of the formula-fed infants from the WHO median are greater in the new infant formula compared with the standard infant formula group. In addition, the observed distributions of anthropometric z-scores should be further used to contextualize the direct comparisons between the intervention groups, as mandated by FDA. For example, 21 CFR § 106.96(b)(4) states that even with no significant differences in weight between the test and control formula-fed groups, it is essential to highlight the degree to which both groups had higher (or lower) weights than expected for age based on the human milk–fed growth standard.

Current FDA regulations (21 CFR § 106.96(b)) do not require that the GMS compare the growth of infants fed the new formula (test group) to

a group of human milk–fed infants. While FDA recognized that potential confounding prevented use of a human milk–fed group as a control, FDA wrote, “the Agency would not object, however, if breastfed infants from the same population as the infants consuming the infant formula under evaluation were included as a concurrent cohort group” (79 FR 7934 at 8010). Regulatory agencies outside of the United States have issued guidance on use of a human milk control group.

A 2021 Health Canada guidance document specifically advises that a “breastfed group” be included as a reference group, in addition to the concurrent control infant formula–fed group. This guidance aligns with the 2004 IOM report recommendation for “dual control groups—breastfed infants and infants fed the previously approved formula without the new ingredient—in order to ensure a thorough analysis” (Health Canada, 2021; IOM, 2004).

The European Scientific Committee on Food (SCF) wrote, “The gold standard for the evaluation of outcomes in infants fed a modified infant formula should be the comparison with outcomes seen in healthy infants who have been exclusively breastfed” (SCF, 2003) and noted a benefit to studying a concurrent exclusively breastfed group when physiologic outcomes other than growth are of interest. Furthermore, Codex specifies that “other ingredients may be added in order to provide substances ordinarily found in human milk and to ensure that the formulation is suitable as the sole source of nutrition for the infant or to provide other benefits that are similar to outcomes of populations of breastfed babies” (emphasis added) (FAO and WHO, 2023).

The scoping review identified 71 articles that included a human milk comparator group (See Appendix E). It showed that inclusion of a human milk-fed group was more common in studies conducted outside of (62 out of 93 total studies, 67 percent) than in (9 out of 22 total studies, 41 percent) the United States.

6. Formula intake data compared to concurrent control group or scientifically appropriate reference

The 21 CFR § 106.96(b)(5) requires that a GMS “compares the data on formula intake of the test group with a concurrent control group or groups and a scientifically appropriate reference.” FDA’s 1996 proposed rule did not require formula intake data to be recorded and reported; the interim final rule added regulation 21 CFR § 106.96(b)(4) and (b)(5), “wherein the protocol of a well-controlled GMS would require information on infant formula intake for both the test and control groups” (79 FR 7934 at 8009). FDA wrote, “These data are needed to provide fair and meaningful interpretation of the study results and

to demonstrate whether the new formula is able to support normal physical growth.”

The committee assessed how data on formula intake have been collected, reported, and interpreted. Most studies in the scoping review reported some measure of formula intake, more commonly in U.S. (82 percent) than non-U.S. (70 percent) studies. Numerous methods were used to collect these data. The most frequently used method was a diary wherein the caregiver recorded the intake at each feeding or each day, subtracting the volume of formula remaining in the bottle from the initial amount in it (Grant et al., 2005; He et al., 2022). Records of intake were collected over a period of days, often 3 days, prior to a study visit (Ahrens et al., 2018; Alliet et al., 2022; Billeaud et al., 2014; Chen et al., 2023; Fleddermann et al., 2014; He et al., 2022; Inostroza et al., 2014; Koo et al., 2003; Kuehn et al., 2022; Lasekan et al., 2011, 2014; Li et al., 2019; Lönnerdal et al., 2016; Marriage et al., 2015; Meli et al., 2014; Parschat et al., 2021; Petersen et al., 2020). Other approaches to data collection ranged from a 24-hour recall reported during a single study visit (Billeaud et al., 2022; Hedrick et al., 2021; Hoffman et al., 2008, 2019; Jiang et al., 2022; Johnston et al., 2015; Li et al., 2019), to study personnel collecting each prefilled bottle after use and calculating intake by the difference.

The number of data collections during a study varied from monthly (Jaramillo-Ospina, 2022) to calculation of each day’s formula volume. Most often, three or four data collections were reported, usually at monthly intervals. The mean values for intake within a study had wide margins of error; coefficients of variation on any mean intake were typically 20 percent or more. Differences in reported mean intake at particular times of infancy were even greater. Data are most often reported as ml/d, but some studies used kcal/d, ounce/kg, or kcal/kg. When formula intake was reported, almost all studies examined by the committee reported no difference. There were three reports of different formula intake amounts between treatment groups with no difference in weight gain (Fledderman et al., 2014; Kuehn et al., 2022; Timby et al., 2014). One of these tested caloric density as a variable and found that the increased volume compensated almost exactly for the reduced caloric content (Timby et al., 2014). The other two studies fed isocaloric formulas and interpreted the difference in intake to mean greater efficiency of utilization of nutrients (Fledderman et al., 2014; Kuehn et al., 2022).

Given no difference in weight gain, it is questionable whether formula intake data are necessary for a determination that a formula supported normal physical growth. Similarly, an increase in efficiency of utilization could be of some potential value to the infant, but given that weight gain was unaffected, the difference in formula intake would not change the conclusion that the formula supported normal physical growth. In no

instances was formula intake the same but weight gain differed; such results would call into question the ability to support normal physical growth regardless of formula intake.

The committee did not analyze the reported formula intake volumes or conduct a meta-analysis to reach a recommendation for best practices for data collection. The committee is unaware of any meta-analysis on the topic. However, energy intake studies have used more precise methodologies to measure formula intake (e.g., deuterium-enriched water (Butte et al., 1991) and doubly labeled water methods (Matsiko et al., 2020; Salazar et al., 2000).

Growth Monitoring Study Assurances—Other Requirements by 21 CFR § 106.121

The committee further examined each part of 21 CFR § 106.121 that elaborates on requirements for a GMS. It requires a narrative about the study, retention of records, and specific analyses (under CFR § 106.121(a) (3) and (a)(4)). A summary of specific information required follows.

1. Statistical evaluation for all measurements, including group means and standard deviations and measures of statistical significance for all measurements for each feeding group at the beginning of the study and at every point where measurements were made throughout the study

The committee noted that for continuous outcomes for which corresponding baseline measurements were obtained, a widely recommended approach is a cross-sectional comparison of values at trial endline, usually with adjustment for baseline measurements as a covariate in the model (often referred to as “analysis of covariance”, or ANCOVA) (Clifton and Clifton, 2019; Twisk et al., 2018; Van Breukelen, 2006). In the absence of meaningful between-group differences (due to randomization), analyses of endline-minus-baseline changes should yield similar analyses as those based on endline measurements (with or without adjustment for baseline), but even minor baseline group differences can render gains/increments subject to regression to the mean artifacts if the analysis does not include baseline adjustment (Twisk et al., 2018). FDA regulation only refers to comparison of means rather than other summary measures, such as the proportion of infants in the lower or upper tails of the anthropometric parameter distributions.

A limitation of the current regulations is that while FDA provides informal guidance, these do not require that manufacturers design the GMS as a non-inferiority (or equivalence) trial, rather than a superiority trial, for anthropometric outcomes. This decision has substantial

implications for sample size and power considerations and data analysis and interpretation. Health Canada and EFSA explicitly recommend non-inferiority designs. With any concern that a new formula may lead to excess rather than inadequate weight gain, a non-inferiority margin could be specified as a maximal weight excess or an equivalence design (two-sided comparison) used so that the GMS is powered to examine differences versus the control in either direction. FDA provides the pharmaceutical industry with guidance regarding non-inferiority trial design and analysis, including the pitfalls of intention-to-treat analyses (FDA, 2016). If per-protocol analyses are conducted, careful attention must be paid to study attrition and missing data.

2. Calculations of the statistical power of the study before initiation and at completion

The GMS sample size and power calculation requirements per 21 CFR § 106.121(a)(3)(ii) include reporting to FDA on “calculations of the statistical power of the study before study initiation and at study completion.” The accepted standard is a 3g/day difference in weight gain compared to the control group for the novel formula, although this is arbitrary, not based on data with any impact on clinical outcomes. The regulations omit critical details of the requirement, including using intermittent time points and interpreting excess weight gain relative to human milk–fed babies. In addition, the regulatory requirement relies on weight only in evaluating meaningful outcomes. Data are also collected on length and head circumference, which are more likely to be associated with protein accretion and developmental outcome than weight is in premature infants (Belfort and Ramel, 2019; Salas et al., 2023).

The Infant Formula Act of 1980, the statute implemented in the regulations, does not set a specific energy requirement for infant formula. Establishing sample size based on the outcome measure of 3g/day difference is not readily done and is arbitrary. Some European formulas go below 20kcal/oz. Although not specified in regulation, the energy requirements for infant formulas provided through the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) must be 20kcal/oz.

Additional challenges and limitations of current requirements include the lack of mention in the regulations about superiority versus noninferiority designs or establishing a priori a non-inferiority margin. In addition, the justification for requiring post-hoc power calculations is unclear and not a conventional practice.

3. Report on attrition and on all occurrences of adverse events during the study

The GMS attrition and adverse events (AEs) requirements per 21 CFR § 106.121(a)(4) include

1. Identification of the infant by subject number and feeding group and a complete description of the AE, including comparisons of the frequency and nature of occurrence in each feeding group and information on the health of the infant during the course of the study, including the occurrence and duration of any illness;
2. A clinical assessment by a health care provider of the infant’s health during each suspected AE; and
3. A list of all individuals who did not complete the study, including the subject number and the reason that they did not complete the study.

Determining if the AE was related to the infant formula consumed involves expert judgment. Most studies identified in the scoping review (>90 percent) reported attrition and AE. Table 4-1 summarizes studies identified by the scoping review that reported on any of these regulatory requirements.

Challenges and limitations of the current requirements include that AE reporting does not use a risk-based approach to determine whether more stringent requirements should be applied based on the risk that the new formula will fail to promote normal growth and development (e.g., novel plant-based protein sources).

Conclusion 4: The committee was unable to identify empirical evidence that supported or refuted the appropriateness of the current requirements as specified in 21 CFR § 106.96(b) and 106.121 or that would support the suitability of alternative designs for evaluation of normal physical growth. It is possible that deviations from these regulations could be scientifically justified and therefore acceptable to the FDA as alternate methods or designs; as well, acceptance of alternative designs may have the additional benefit of contributing to international harmonization of regulatory statutes. Specifically, the committee was not able to establish information to support the necessity of any specific number of anthropometric data collection times or schedule of those times during the course of a growth monitoring study. This was also the case for supporting a mandatory enrollment age of less than 2 weeks of age or a 15-week trial duration. Other mandated data, such as formula intake and anthropometric measurements other than weight, were not clearly identified as being needed in all growth studies. Decentralized designs whereby anthropometric data collection is performed by parents/caregivers or health professionals who

TABLE 4-1 Number of Studies Conducted Within and Outside of the United States Reporting the Food and Drug Administration Regulatory Requirements for Growth Monitoring Studies for Infant Formulas Based on Published Studies in the Scoping Review

NOTES: AEs = adverse events; FDA = Food and Drug Administration; n = number; NR = not reported.

* Data added by the committee

FDA-required variables are those described in 21 CFR § 106.96 and 21 CFR § 106.121.

SOURCE: Scoping review commissioned from the Academy of Nutrition and Dietetics (see Appendix D).

are not study personnel may be an alternative method if suitable efforts are made to ensure that non-study personnel have met minimum standards of precision and accuracy in the conduct of the measurements.

Recommendation 4: The Food and Drug Administration (FDA) should publish a single guidance document that (1) describes the preferred design features of a growth monitoring protocol and explains how FDA uses required information in its evaluation that a formula supports normal physical growth and the conditions under which alternative designs may be acceptable to FDA; and (2) outlines the conditions under which a growth monitoring study is needed. That guidance should take into account (1) whether the change in infant formula could reasonably affect growth, (2) if a new ingredient is normally found in human milk, (3) the extent to which prior studies have examined the effect of a new ingredient on growth, and (4) information about the effects of addition of the ingredient on the level of or bioavailability of a nutrient whose deficiency over the course of the study would be manifested in reduced growth.

Conclusion 5: Current FDA regulations do not clearly define the nature of a major change to an infant formula that requires a growth monitoring study (GMS) other than the requirement that all ingredients in infant formula must be shown to be safe and suitable for their use in formula (21 CFR § 106.40). Guidance is lacking on the specific type of evidence required under various conditions of change in infant formula such as potential effects of a change in formula composition or processing on infant growth; if a new ingredient is normally found in human milk; the extent to which prior studies have examined the effect of a new ingredient on growth; and information about the effects of addition of the ingredient on the level of or bioavailability of a nutrient whose deficiency over the course of the study would be manifested in reduced growth.

Conclusion 6: Weight gain is conventionally the primary outcome variable used to assess normal physical growth. However, there is a lack of strong evidence to support the commonly implemented, but not described in regulation, usage of a 3g/day maximum allowable difference in weight gain between test and control groups. Furthermore, there is no statistical justification to prioritize the analysis of average weight gain over the analysis of endline weight as the primary outcome in the context of a randomized controlled trial. Estimation of differences in endline mean z-scores when comparing test and control groups, or when

comparing any group to the WHO/CDC growth standard, is a reasonable alternative to comparisons based on differences in weight gain. The establishment of a standard acceptable non-inferiority or equivalence margin (e.g., ±0.5 z-scores) could facilitate international harmonization of regulatory requirements.

ALTERNATIVE APPROACHES

Complementary Outcome Measure of Infant Growth—Body Composition

Anthropometric measures, predominantly weight velocity, body mass index (BMI), weight z-scores, or weight-for-age or length z-scores, are used most often in GMSs, as such measures are relatively easy and inexpensive to assess or calculate and efficient for larger sample sizes. Limitations include interobserver variations and the positioning of infants on length boards due to movement. A major drawback with anthropometric measures is that they do not directly measure fat or lean mass. The use of BMI can lead to both under- and overestimation of body fat as it cannot distinguish body composition (i.e., fat versus lean mass) (Rothman, 2008). At present, there are no published BMI z-scores under the age of 2 years (CDC, 2024b).

The idea of measures of body composition as a “quality factor” in assessing infant formula dates to the 1988 CON report on types of clinical studies to assess the nutritional safety and suitability for healthy term infants that should be performed before marketing infant formula (AAP, 1988). In addition, the 2004 IOM report on evaluating the safety of ingredients intended for use in infant formula recommended that body composition be assessed even while acknowledging that it was difficult to measure accurately, and the best methodology requires expensive equipment (IOM, 2004). However, to the committee’s knowledge, no regulation or guidance document on GMS in formula-fed infants has adopted body composition as a measure of quality. A 2019 National Institutes of Health workshop highlighted the importance of body composition measurements from birth through 5 years, noting the applicability in health care settings, clinical research, and national surveys (Gallagher et al., 2020). As pointed out by Norris et al. (2024), body composition compared with anthropometric measurements may better reflect biological responses to nutrition and environmental exposures.

The scoping review identified that 13 of the 143 studies included measures of fat mass and 12 included measures of lean mass using one or more of the methods to assess body composition in infants. The most frequently reported method to obtain measures of fat and lean mass was measuring body density by air-displacement plethysmography (ADP), usually with

commercial equipment (“PEAPOD”; Infant Body Composition System, Cosmed). Equipment for ADP in infants is not valid beyond about 6 months of age or 8 kilograms in weight (Rosendale and Bartok, 2012).

Skinfold thickness measures using a caliper designed for infants at one to four body sites (subscapular, triceps, supra-iliac, biceps) was also employed to quantitatively measure fat mass. While this method produces an indirect measurement, it is efficient and effective. However, limitations include that the percentage of fat mass is often derived using the Slaughter equation formulated for fatness predictions in children rather than infants (Slaughter et al., 1988); high interobserver variations; low reproducibility; and unreliability for measuring intra-abdominal fat or central obesity, although this is likely not an issue in very young infants (Fattah et al., 2009; Marshall et al., 2016).

Dual-energy x-ray absorptiometry (DXA), which is the most accurate direct measure of fat mass and considered the “gold standard” for body composition in infants (Demerath and Fields, 2014; Pietrobelli et al., 1996), was used in three studies identified by the scoping review (Inostroza et al., 2014; Martin et al., 2014; Putet et al., 2016). However, as pointed out in the National Institutes of Health workshop summary, it faces the challenge of movement artifact, a problem inherent in young active infants (Gallagher et al., 2020). The doubly labeled water method can also provide measures of total energy expenditure but requires water labeled with two stable isotopes to be consumed and urine samples collected (Escribano et al., 2012).

In all studies in the scoping review, body composition was reported in absolute amounts of percent body fat, total body fat (fat mass in kg multiplied by percent fat mass), fat mass index (kg/m² divided by height (m) squared), and total lean mass or fat-free (lean) mass index. Measures were taken as early as 1 month of age and up to 12+ months. The findings of the 13 studies that conducted measures of body composition were inconsistent in interpretation. The most common was that while fat and lean measures did not differ between experimental and control infant formula groups, the percent body fat and lean mass of formula-fed infants often did (sometimes higher and sometimes lower) compared to a human milk–fed comparison group. In two studies in which infants were fed infant formula of different protein content (Liotto et al., 2018) or variations in bovine milk fat globule membrane concentration (Jaramillo-Ospina et al., 2022), fat-free mass was higher in formula groups compared to the concurrent human milk–fed controls.

Based on the committee’s review, measures of fat and lean mass parameters in response to alterations in feeding (e.g., protein or energy concentration or quality, bioactive factors) in infancy appear promising to provide insights into the quality of growth. Limitations to widespread use of such measures include shortcomings for ease of use in infants and

the cost of the equipment (Ceniccola et al., 2019). ADP and hydrometry (deuterium oxide dilution) are more accurate and precise, as both methods directly measure fat mass, but sensitive to shifts in total body water (TBW) changes in early infancy (Fattah et al., 2009). However, both methods are time consuming and not suitable for large-scale studies (Kuriyan, 2018).

Until recently, a major limitation in application of body composition measures in infants has been the absence of normative data (Hills et al., 2023), as no reference standards were available by age and sex. At its open session workshop in July 2024 (NASEM, 2024b), the committee was apprised that ADP had recently developed such standards for TBW to age 6 months in a multiethnic sample (Africa, Asia, Oceania, and South America) in nearly 4,000 predominantly human milk–fed infants (Murphy-Alford et al., 2023; Norris et al., 2024). Specifically, sex-specific reference charts were published for fat mass index, fat-free mass index, and percent fat mass for 470 infants ages 0–6 months (Murphy-Alford et al., 2023). Other reference standards for fat mass by ADP have been reported across four studies in three countries (United States, Ireland, and Italy) that provide percentile curves for the same fat and fat-free measures (Norris et al., 2019). Once such reference data are widely available and align with standards of growth measures, such as the WHO/CDC standard growth curves, measures of body composition in infants may provide more biologically relevant information than standard anthropometric data.

Study Design

FDA has long recognized that different types of changes to the composition or processing of an infant formula do not require the same type of supportive evidence to establish safety and nutritional suitability. In 1985, FDA issued guidelines on testing formulas that distinguished major from minor changes (see Box 4-1): “FDA has recognized that premarket clinical evaluation in humans may be appropriate whenever certain changes affecting the nutritional profile of an infant formula are made, particularly in the case of new or reformulated products. FDA has also recognized that the degree and complexity of clinical testing needed will vary according to the extent of the changes in the formula” (FDA, 1985). The definition of major change was incorporated into the Infant Formula Act by reference in its 1986 amendment.

However, FDA’s 2014 final rule on quality factors established only one type of clinical study, GMS, for the premarket evaluation of a new or reformulated infant formula. As described earlier in the chapter, the GMS requirements are stringent. Other bodies have recommended a tiered approach to assessment of safety of new formulas (IOM, 2004), where initial clinical assessments might identify areas for more detailed study.

Such a tiered approach is used by FDA to determine study requirements in other areas, such as toxicology (FDA, 2000).

The committee noted that a risk-based model might be applied to clinical assessment of new or reformulated infant formulas and considered not only alternative clinical study designs but situations in which animal testing or analytical data could provide sufficient evidence on which to conclude about safety and nutritional adequacy. The committee recognized that normal physical growth is the default measure established by FDA for normal physical growth and that weight gain is the most important outcome to assess normal physical growth. However, as FDA stated in its 1985 guidelines, the degree and complexity of clinical testing needed varies, and the nature of evidence needed to support the conclusion of normal physical growth depends on the extent of the difference from existing formulas in composition or processing of a new or reformulated infant formula, and the extent to which information is available on similar manufacturing or composition. RCTs are much more complex and costly than when FDA first proposed the requirements for a GMS in 1996 (Zabor et al., 2020). The committee heard from invited

experts during an open session that questions raised by FDA on the quality factor of normal physical growth were usually about operational aspects of the study that did not affect the conclusions on the primary outcome (NASEM, 2024c). Efforts to collect the types of data established to be necessary to assess normal physical growth are needed.

Major changes, as defined by 21 CFR § 106.3, do not have equal potential to affect growth. The nutrients required in infant formula reflect the essential nutrients needed to support infant growth and development; levels are largely based on the concentrations of the nutrients in human milk, set as an estimate of adequate intake found in the Dietary Reference Intakes (Grote et al., 2011; Lönnerdal, 2014). All major changes listed are not presumed to enhance growth; they are considered major changes because they may have a significant adverse impact on nutrient levels or bioavailability, which underlies FDA guidance to test for non-inferiority of growth (weight gain or weight velocity) in GMS. In practice, not all types of major changes require GMS.

For 21 CFR § 106.3, examples are provided on what constitutes a major change (see Box 4-1). Example 1 is directly relevant to FDA’s enforcement discretion exercised during the 2022 infant formula shortage that originated in the United States. While these formulas are coming into compliance with existing regulations under a process laid out by FDA, its acceptance assumed, without the evidence of a GMS, that the formula would support normal growth. Example 2 is exempted by the current regulations that require the GMS to be conducted using the form most vulnerable to nutrient losses in processing (i.e., liquid) (21 CFR § 106.96(c) (2)(iii)). Examples 4 and 6 include different methods of processing of the same formula that do not typically require a new GMS but might require evidence of comparability of the nutrient levels or an animal study demonstrating comparable growth. Example 7 is exempted by 21 CFR § 106.96(c)(i). Consequently, the most common reason to conduct a GMS is for either a new macronutrient source (Example 3) or a new constituent not required by regulation (Example 5). This is borne out by the studies captured in the scoping review; virtually every study was undertaken because of the types of major change described in Examples 3 and 5.

Most studies (108 of 135) in the scoping review that reported infant weight data were RCTs, evidence that this study design is ordinarily used. Of the remaining studies, eight were meta-analyses. The scoping review found no apparent relationship between frequency of a report of a significant difference in a weight measurement and design of a trial as an RCT.

Given that no GMS is required for many major changes in infant formula compositions, it is plausible that alternative methods to an RCT might satisfactorily document normal physical growth when a GMS is needed. The committee recognizes the lack of empirical evidence to

support a risk-based approach; however, the evidence from the scoping review suggests that evaluation of normal physical growth is not sensitive to modifications of a formula that, by regulation, satisfies all known nutrient requirements for infant growth. If growth (weight gain) is the primary outcome, comparison to standards or historical controls may be sufficient; when substances are known to not affect growth, the effect of including them in a particular formula matrix processed in a particular manner could be studied in preclinical models. One irreplaceable attribute of an RCT is comparing AEs in infants fed a new versus a commercial formula, where there is no standard of reported AEs. Another attribute is to demonstrate a physiologic benefit to health and development of the infant that might occur without any effect on growth.

An RCT may not always be needed, particularly when the representativeness of the study group to the intended-use population is established, data collection is unbiased, and safety measures can be compared to population references. Table 4-2 illustrates alternative study designs identified by the committee.

Collection of Growth Data in a Trial Where Growth Is Not the Primary Outcome Measure

While normal physical growth must be established to ensure new or modified infant formulas are safe, many studies on new ingredients are undertaken to examine physiological or biochemical effects. These are efficacy studies, so they are ordinarily double-blind RCTs. They may also obtain growth data, even though no effect on growth is expected. The primary outcome is not anthropometric data specified in GMS; they may be immune, neurologic, or visual, for example. When growth is not the primary outcome, the requirements of measuring normal physical growth are not driving study design, and parameters of data collection on growth measures may be given flexibility appropriate for a secondary outcome measure. Given no empirical evidence for most of the specific regulatory requirements of a GMS, one alternative study design would be to collect the evidence on which regulatory decisions are made as a secondary outcome.

Other regulatory requirements, such as enrollment at less than 2 weeks of age, data collection at six specified times of measure, and the manner of comparing groups or plotting data for individuals, are not critical to the study and may not be necessary to reach a conclusion on growth, primarily assessed by weight gain. Anthropometric data still would be collected over 4 months (an undisputed duration) and plotted in real time during the study as a safety measure. Growth would still be assessed over the course of the study and compared to both the WHO/CDC growth standard and

TABLE 4-2 Alternative Study Designs for Consideration to Demonstrate That a Formula Supports Normal Physical Growth

NOTES: AE = adverse event; FDA = U.S. Food and Drug Administration; GMS = growth monitoring study; HTST = high temperature, short time.

^a Variations on the RCT as laid out under current regulations should be acceptable: all-comers study (randomized into test vs. control arms and then stratified on day 14 into three groups within each arm; strata = no formula, mixed formula with breast milk, only formula); unbalanced RCT with fewer subjects in the control group than the test group. Study design features, such as enrollment by a particular day of life, duration of observation, extent to which non-study feedings are allowed, recording of formula intake, statistical power to detect particular differences in outcome, manner of collecting and reporting AEs, and manner of reporting of anthropometric data, should all be matched to the documented need for such data to address the outcomes of interest of the study.

concurrent control. Studies would have to be powered correctly to detect a meaningful difference in the secondary outcome of growth.

If there is a possibility that a change in formula composition or processing would affect growth or the study is undertaken to assess a growth outcome, it should be designed to assess weight gain. The details of existing regulations are for that purpose and would be further informed by guidance from FDA on the conduct of GMS.

Conclusion 7: Conducting a research study in which a new formula is compared to an existing approved one, referred to by the term “concurrent control” as used in 21 CFR § 106.96(b)(5), is conventionally interpreted by investigators and FDA to mean the need for a randomized controlled trial (RCT). An RCT may be needed to demonstrate the absence of a negative effect on growth of infant formula–fed infants because of a change in formulation or processing of an infant formula. However, an RCT may not be needed under certain conditions, and suitable data could be generated in a single-arm study in which the growth of infants receiving the test formula is compared to the WHO/CDC growth standard.

REFERENCES

This page intentionally left blank.