Infant Mortality Decline in the Late 19th and Early 20th Centuries:The Role of Market Milk
Starting in the late 19th and early 20th centuries, a precipitous decline in infant mortality was observed in the United States. Economic growth, improved nutrition, new sanitary measures, and advances in knowledge about infant care all contributed to this decline in infant mortality. Little is known, however, about how these individual factors affected disease-specific components of infant mortality over time. Systematic review of historical data suggests that cleaning the market milk supply was the single most important contributor to this decline in both diarrheal and overall infant mortality, and that this development played a far more important role than family income, other sanitary measures, or medical intervention.
Beginning in the early 20th century, a rapid decline in infant mortality was observed in the United States and Western Europe. This decline resulted mainly from a progressive reduction in mortality from infectious diseases, notably diarrheal diseases (Beaver 1973; Bideau, Desjardins, and Brignoli 1997). This decline has been broadly attributed to economic growth, improved nutrition, sanitary [End Page 585] measures, and advancing knowledge in infant care (McKeown 1976; Preston and Haines 1991). However, there have been few investigations of how these individual factors affected disease-specific components of infant mortality over time, particularly for diarrhea (Beaver 1973; Condran and Cheney 1982; Condran and Lentzner 2004). One major impediment to such investigation arises from a lack of longitudinal historic data that would enable us to examine all factors simultaneously over time.
This paper attempts systematically to review and reanalyze pertinent historic data and to estimate the impact of diarrhea on the decline in infant mortality in the late 19th and the early 20th centuries. Since adequate longitudinal data on several factors is no available, various fragmentary cross-sectional data on different populations in different time periods have been thematically structured to construe how these factors (income, sanitation, infant feeding practice, and quality of market milk) would have affected diarrheal and overall infant mortality in this period.
Infant Mortality in the 19th and 20th Centuries
During the second half of the 19th century, infant mortality in different regions of the United States and Western European countries ranged from about 100 to 350 per 1000 live births, but in each region remained roughly constant (Alter 1997; Mitchell 1992; Preston and Haines 1991; U.S. Census Office 1870–2004). In this period, a major portion of deaths were those of children; in the United States, approximately one out of every three deaths was a child under five years of age and one out of every five deaths was an infant. A sharp and continuous decline began in the first decade of the 20th century (Figure 1). In the United States, over 50% of the total decline in infant mortality during the 20th century occurred in the first three decades.
In the United States between 1870 and 1900, roughly 40% to 50% of all reported infant deaths were from infectious diseases. Diarrhea was responsible for about half of all infectious disease deaths in infancy. As a cause of death, infantile diarrhea was listed as summer diarrhea, intestinal catarrh, cholera infantum, enteritis, or enterocolitis. The other major cause of infant death was respiratory infection (bronchopneumonia and pneumonia), causing about 17% to 36% of all infectious disease deaths. Communicable diseases such as measles, scarlet fever, whooping cough, diphtheria, croup, and smallpox were responsible for 20% to 29% of all infectious disease deaths. Typhoid fever, cholera, and parasitic enterocolitis were rather infrequent causes of infant deaths. [End Page 586]
Artificial Feeding and Diarrheal Death
High mortality in infants who were fed cow's milk instead of breast milk was reported in Norway and Sweden in 1825 (North 1921). But it was during the latter half of the 19th century that artificial feeding attracted increasing attention as a cause of infant death. During the siege of Paris, 1870–71, while the general mortality doubled, infant mortality fell 40%. When the milk supply failed with the siege, the Parisian women nursed their children more (Newman 1906). The American Civil War broke out in 1861, and in the following year the import of cotton to Great Britain fell. During this cotton famine in Lancashire, the general death rate increased but infant mortality declined, while mothers were not at work at the mills (Newsholme 1899). In Germany during the deprivation of the First World War, breastfeeding increased and again, infant mortality declined (Hecker 1923; Seidmayer 1937).
In Paris in 1897, among 2,840 infant deaths, 1,470 (51.7%) were from diarrhea, and out of these infants with diarrheal death, more than 90% were artificially fed (Budin 1906). In Bavarian villages and cities in 1878 to 1882, infant mortality varied from 174 to 372 per 1,000 live births (Knodel and Van de Walle 1967). Examination of this tabulated data shows that more than 60% of this variation was attributable to the difference in prevalence of breastfeeding. Newsholme (1906) studied diarrheal deaths in Brighton, England, for three years, from 1900 to 1902. Artificially fed infants were 10 times more likely to die from diarrhea than breastfed infants in the first year of life. And among all diarrheal deaths, 91% of them were not breastfed. [End Page 587]
In 1911, a Boston survey showed overall infant mortality of 127 per 1,000 and diarrhea was responsible for 29% of all infant deaths (Davis 1913). In this survey of infants between two weeks and one year of age, the artificially fed died from diarrhea14 times more frequently than the breastfed. Between 1912 and 1915, the U.S. Children's Bureau (1913–1923) conducted field surveys on infant mortality in eight cities. Analysis of this historic data shows that about half of the variation in infant mortality among these cities could be accounted for by the differences in their breastfeeding practice. In the Baltimore survey, compared to breastfed infants, artificially fed infants had higher infant mortality from gastro-intestinal diseases at every month up to the 10th month of life. It was notable between the fourth and eighth months of life that exclusively artificially fed infants had about 26 times greater mortality from gastrointestinal diseases than did infants who were exclusively breastfed.
Cow's Milk and Infectious Etiology of Infantile Diarrhea
It was in the 19th century that diarrhea was recognized as the main cause of infant mortality, with its summer peak mostly consisting of diarrheal deaths (Budin 1906; Combe and Bell 1840). This seasonal pattern of infant mortality, particularly in large cities in the United States, had gradually abated with a decrease in diarrheal mortality over several decades from the late 19th to the early 20th century (Condran and Lentzner 2004).
Hot weather was initially blamed as a direct cause for high infant mortality in summer, particularly for diarrheal diseases. However, this notion gradually faded, when with the emergence of the germ theory in the late 19th century, increasing evidence began to show that contaminated cow's milk with a large number of bacteria was closely associated with diarrheal deaths. By the first decade of the 20th century, it was apparent that the cow's milk could be a vehicle for epidemics of contagious diseases such as typhoid fever, cholera, dysentery, diphtheria, scarlet fever, and septic sore throat (Parker 1917; Rosenau 1912). However, these reported epidemics were sporadic in nature, not seasonal, and often limited to a small number of people including adults as well as children. Thus, these epidemics were different from typical infantile diarrhea.
Sedgwick and Batchelder (1892) found that, with moderate precautions on the part of the milker, the number of bacteria in fresh milk did not exceed 500 to 1,000 per cubic centimeter. Park (1901) showed an exponential growth of bacteria in milk within a day, attaining billions of bacteria per cubic centimeter when temperature was kept at 86°F or higher. At the beginning of the 20th century, bacterial counts of market milk supply in six U.S. cities ranged from 1.4 to 9.4 million per cubic centimeter (Park 1901; Health and Sanitation Committee 1904). This high bacterial count in milk was similar to that in sewage in several cities at that time (Rosenau 1912). For two summer months in 1901–2, Park and [End Page 588] Holt followed healthy infants living in tenement houses in New York City. In the 10-week observation period, 41 of 390 artificially fed infants died, while none of 31 breastfed infants died (Park and Holt 1903). Milk fed to these children in New York City showed 239 varieties of bacteria. None of these bacterial species seemed to have any consistent relationship to infantile diarrhea. Other subsequent studies similarly found multiple bacterial species in commercial cow's milk, but failed to link any of these to diarrhea (Meckel 1990). Early researchers were frustrated by their inability to fulfill Koch's postulates, since they were after a single bacterial pathogen, as previously discovered for typhoid fever, cholera, or dysentery. Instead they found a large number of different enteric bacterial species both in market milk and in stools of infants with diarrhea, but also in stools of infants without diarrhea.
Thus, in the first several decades of the 20th century, it was not possible to relate any specific bacteria in milk to infantile diarrhea, but the bacterial count of milk was widely used as a tool for monitoring its quality (Meckel 1990). Milk reformers justified this monitoring with the following reasons. First, dirty cows and barns produced milk with high bacterial counts. Second, milk delivered from a far distance showed higher bacterial counts, particularly with improper cooling during transit or a delay in delivery. Third, infants developed diarrhea more frequently when fed milk with high bacterial counts than milk with low bacterial counts. Therefore, a high bacterial count in milk consequently became a marker for unsanitary, poor-quality milk.
Current concepts for the infectious etiology of infantile diarrhea began to emerge in the 1940s when diarrhea was no longer a major cause of infant mortality. It should be noted, however, that this idea was already suggested in a study by Lasage in 1897 (cited in Newman 1906). Lasage isolated Bacillus coli (E. coli) from stools of children with diarrhea. The sera of 40 infants with diarrhea agglutinated samples of Bacillus coli from 39 other children with diarrhea. In 1912, Bahr and Thompsen demonstrated that the coli isolated from newborn calves suffering from diarrhea had the same fermentation reaction as those coli bacteria recovered from infants with diarrhea. Subsequently, Adam (1923, 1927) found that certain strains of colon bacilli caused infantile diarrhea and that these toxic strains possessed a fermentation pattern that was different from nonpathogenic strains of the same species. However, the significance of all these observations was totally obscure until the 1940s. Bray (1945) and Bray and Beavan (1948) studied bottle-fed infants with severe diarrhea. They produced a rabbit serum against a strain of E. coli that was isolated from one of the infants with diarrhea. They then showed that 42 of the 44 E. coli strains obtained from infants with diarrhea were agglutinable, while only four such strains were found among 100 infants without diarrhea. This work led finally to the acceptance of a bacterial etiology of infantile diarrhea. Subsequent work by others have shown that, in addition to enteropathogenic E. coli, other pathogenic bacteria and also a number of viral species can cause infantile diarrhea. [End Page 589]
Cleaning the Market Milk Supply
In the United States, cleaning the market milk supply took a century. In the first period, 1840 to 1910, the major effort was to maintain compositional integrity of milk. In the late 1830s, swill milk produced from cows fed distillery waste accounted for more than half of all market milk in large northeastern cities (Shaftel 1978). Campaigns against swill milk started in the 1840s, but actual prohibition of swill milk began two decades later in the 1860s, following a public outcry and multiple investigations (Leavitt 1982). Also, during this period much of market milk was skimmed of its cream and adulterated with added water, molasses, chalk, baking soda, or other chemicals (Meckel 1990). Beginning in the early 1880s, anti-food adulteration bills were passed by a number of states, and these were in effect anti-milk adulteration bills. In the early 1890s, many states established the standard for market milk quality, 12% solid, of which 3% had to be fat (Alvord and Pearson 1903). Statistics on the quality of milk supply in Philadelphia indicated that between 1889 and 1893, an average of 8.5% of inspected milk was skimmed or watered. By 1903, the percentage had dropped to 1.5%, and it was less than 1% by the second decade of the 20th century (Condran, Williams, and Cheney 1984).
In the second period, 1870 to 1940, a vigorous public health movement to prevent the bacterial contamination of milk was launched. In this period, the market milk supply gradually became safer, with improved sanitary conditions in dairy farms, pasteurization of milk, keeping milk at low temperatures during shipping and delivery, and prohibition of the sale of loose milk (unpackaged bulk milk stored in a large canister and sold using a dipper) in grocery stores.
In 1887, Henry Coit, a pediatrician who had lost his infant son to contaminated milk, embarked on a lifelong campaign to make milk safe, and his efforts resulted in a certification program for milk safety (North 1921). High-quality non-pasteurized milk was produced under the supervision of a medical milk commission that, in minute detail, regulated handling of cows, dairy environments, standards for cooling and shipping milk, and bacterial counts of milk (American Association of Medical Milk Commission 1912). Unfortunately, use of this certified milk did not spread widely, primarily because of its higher retail price (Fuchs and Frank 1938).
In the United States, commercial pasteurization was introduced in Baltimore in 1891 and then spread to other cities. An ordinance requiring the pasteurization of milk was passed in Chicago in 1908, then in New York City and Philadelphia in 1912, and subsequently by other large cities (Fuchs and Frank 1938; Lederle and Raynor 1912). In 1912, New York City became the first city to introduce a grading system (Grades A, B, and C) for market milk. The grading was mainly based on the bacterial count of milk and sanitary condition of the dairy (Brown 1916). By 1920, most large cities in the United States required both the pasteurization of milk and grading (Meckel 1990). However, this [End Page 590] change was slow to spread beyond the large cities. In 1936, the U.S. Public Health Service surveyed milk supplies in urban communities of over 1,000 population (Fuchs and Frank 1938). Of a total of 6,488 communities, 2,654 (40.9%) responded to the survey. In these communities, non-pasteurized raw milk still constituted 27% of all market milk. And of 2,312 communities, only 53% graded their market milk.
Another significant development in the milk reform movement at the turn of the 20th century was a worldwide campaign for the provision of safe milk to infants in low income families. The first milk depot opened in St. Gertrude District in Hamburg, Germany in 1889 (North 1921). In the same year, Dr. Henry Koplik established a milk depot in Good Samaritan Hospital in New York City. These were followed by the Goutte de Lait at the Charité Hospital in Paris by Budin in 1892, and the first large-scale milk station in New York by a philanthropist, Nathan Strauss, in 1893. Strauss was an ardent campaigner and responsible for the opening of pasteurized milk stations in many cities in the United States and Europe (McCleary 1905; Straus 1917; U.S. Children's Bureau 1913). Milk stations gave away, or sold at a reduced price, millions of bottles of clean and nutritionally sound milk and taught mothers how to take care of their children (Goodwin 1916).
In 1911, the New York City Health Department introduced a citywide milk station program for low-income families (Van Ingen and Taylor 1912). Following the introduction of this program, summer infant mortality in the city from June through October declined from 59.1 per 1,000 to 47.5 per 1,000 in one year. For the infant group enrolled to the milk station program, summer infant mortality was quite low, 26.9 per 1,000. Reanalysis of the tabulated data on mortality by district shows that there was a significant correlation between the proportion of infants participating in the program and that district's summer infant mortality: the higher the proportion of infants enrolled in the program, the lower was the district's summer infant mortality (r = 0.64).
However, these milk stations in large cities supplied only an infinitesimal fraction of the milk consumed by the nation's urban infants. Milk reformers conceded that even if such stations doubled or tripled in number, the problem would not be solved, unless the poor quality milk could be weeded out from the entire urban milk supply (Meckel 1990).
Water Purification and Diarrheal Mortality
Water directly given to an infant or used to dilute cow's milk might have been contaminated and caused diarrheal disease. But it would have not been an important vehicle for infantile diarrhea. For bacterial growth, water is a far less favorable medium than milk. Moreover, a major portion of the body's water need in infancy is supplied by milk, not by drinking water. Preston and van de [End Page 591] Walle (1978) examined mortality rates in the three largest urban areas of France in the 19th century and found a progressive decline in mortality, starting around mid-19th century and continuing into the first decade of the 20th century. They suggested that improved water supply and sewage disposal, probably in conjunction with improved nutrition, reduced the incidence of diarrheal disease and other infectious diseases. Although the decline in mortality was most remarkable for those below age 15 or 20, little advance was made in infant mortality. Condran and her associates (1984) provided the most convincing evidence against water as an etiologic vehicle for infantile diarrhea. In Philadelphia, filtration of the water supply began one water district at one time, starting in 1902 and extending throughout the city by 1909. Typhoid mortality dramatically declined in one district after another, coinciding with the establishment of water filtration, but despite this decline, the diarrheal mortality of children under two years of age did not change significantly in any water districts throughout this period, or in the subsequent decade.
Socioeconomic Factors, Sanitation, and Diarrheal Mortality
Although there is ample evidence that diarrheal mortality was mostly attributable to artificial feeding, particularly feeding poor-quality milk, inappropriate maternal care of infants was often blamed. Infant mortality from both diarrheal and non-diarrheal diseases was high in infants of immigrant mothers, in tenement houses with poor sanitary conditions, and in families with lower income (Hibbs 1916; Woodbury 1922).
Davis (1913) examined infant deaths in Boston for the year 1911. Reexamination of his data reveals that compared to breastfed infants, bottle-fed infants had a risk of infant death from diarrhea 13.5 times higher, and a risk of infant death from non-diarrheal causes was 4.8 times higher. After adjusting for the difference in feeding method, only infants of Italian-born mothers had a slightly higher risk of diarrheal infant death compared to infants of native mothers.
Reanalysis of the Baltimore survey (U.S. Children's Bureau 1923) showed that mortality risk of infants of the foreign-born mothers as a whole was not higher than infants of native mothers, both with and without adjustment of father's earnings and feeding method. However, both father's earnings and feeding method were independently related factors to infant mortality. The role of father's earnings in the mortality disparity might have simply reflected various factors often associated with poverty, such as poor sanitary and housing conditions, mother's employment outside of the home, and mother's ignorance in infant care. To explore this question, aggregated tabulated data in the Baltimore survey have been examined, since individual infant data are not available. The risk of infant death was examined by ward, with father's earnings, percent of families with bath in house, and percent of houses connected to sewer. Only [End Page 592] father's earnings were associated with the risk of infant death. Using other tabulated data on the father's earnings in this report, infant mortality was regressed by father's earnings, percent of infants breastfed at sixth month, crowding in the house (persons per room), and mother's employment away from home. Infant mortality was related to breastfeeding at six months and to crowding, but it was not significantly affected by father's earnings or mother's employment outside the home. The effect of breastfeeding was greater than that of crowding. The disparity in mortality by father's earnings was mainly attributable to a far greater difference in mortality among artificially fed infants than among breastfed infants (Figure 2). The breastfed infant group with the lowest father's earnings had a lower risk of death throughout the first year than did the artificially fed group with the highest father's earnings.
Interpretation of results from the Baltimore survey is somewhat speculative, since the analysis is based on the aggregated data and not on individual infants. Nevertheless, these results support the contention that the feeding method and the father's earnings were two major factors in infant mortality. In their relationship to infant mortality, feeding method had a far greater effect than family income. This contention concurs with observations made by contemporary physicians. In 1906, Knox acknowledged that in Baltimore more than nine-tenths of all deaths from summer diarrhea occurred in artificially fed babies; but such cases were of the utmost rarity among breastfed infants, even in the poorest homes (Fee and Brown 1999). In their textbook on nutrition in early childhood, McLean and Fales (1925) stated: "The infant mortality rate among certain groups of the poor who live in most unhygienic surroundings but who habitually nurse [End Page 593] their offspring has been found to be markedly lower than that among even better classes who resort mainly to artificial feeding" (p. 59).
How, then, did father's earnings influence infant mortality, particularly mortality due to diarrheal deaths among the artificially fed? In the first several decades of the 20th century, market milk was still of variable quality. Because of its higher price, certified milk had a relatively small market share. Even at the height of its popularity, it represented only the very top layer of the urban milk supply; around 1910, this milk consisted of far less than 1% of all milk sales in Boston, New York, Philadelphia, and other large cities, and its market share remained similar in 1936 (Fuchs and Frank 1938; Rosenau 1912). Park (1901) reported bacterial counts of milk sold in different areas of New York City. In summer, in a poor tenement district, the average bacterial count of market milk was 15 times higher than that in a well-to-do area. A commission appointed by the New York City Department of Health examined use of market milk by 31,507 poor families (Milk Commission 1931). In Manhattan, only 2% of these families used Grade A milk, and the remainder used either Grade B or loose milk. It was not far better in other boroughs of New York City, with about 80% of poor families using the latter two poor-quality milks. In the early 20th century, the price of milk was relatively expensive (U.S. Bureau of the Census 1960, 1975). The retail price of whole milk was 7 cents per quart in 1910, and increased to 13 cents per quart in 1920. A worker's average annual earnings were $575 in 1910 and increased to $1,342 in 1920. In general, the price of top-grade milk was often almost twice as expensive as low-grade milk (Alvord and Pearson 1903; Fuchs and Frank 1938).
These observations suggest that in the early 20th century, family income might have been a mediating factor in diarrheal infant mortality—particularly among the artificially fed—reflecting the family's ability to afford good-quality milk. This contention is supported by the finding that, in 1911, summer mortality was reduced by half in infants—mostly from low-income families—who received good milk from the New York City milk station program (Van Ingen and Taylor 1912).
Breastfed infants were not entirely free from diarrheal deaths, however, and such deaths occurred more frequently in indigent families. In these infants, diarrhea may have been domestically transmitted via a fecal-oral route from an unsanitary environment. Likewise, some of the artificially fed infants would have developed diarrhea by this route, rather than from ingestion of the contaminated market milk. Purchased market milk also might have been contaminated at home. Krasner-Khait's (2000) history of refrigeration tells us that since the late 19th century, a wooden icebox was commonly used in most homes but rarely in the poorest tenements. In 1921, 5,000 mechanical refrigerators were manufactured in the United States. Ten years later, that number grew past 1 million. But mass production of modern refrigerators did not begin in earnest until after World War II. Indigent families would have been more likely to purchase poor-quality milk and [End Page 594] to have difficulty storing it properly. A family's economic status, sanitary conditions, and maternal knowledge of infant care would have been significant interacting factors for infantile diarrhea. The impact of all these domestic factors on diarrheal mortality, however, would have been small, given that artificially fed infants in wealthy families had far higher diarrheal mortality than did breastfed infants in indigent families; the quality of market milk was evidently a more important factor in infantile diarrhea than was the domestic environment.
Urban and Rural Differential in Diarrheal Mortality
In 1890, diarrheal infant mortality was six times higher in urban areas than rural areas. To account for this disparity in diarrheal mortality, there must have been a difference in either one or both of two determining factors for infantile diarrhea: its prevalence, and its case fatality rate. The prevalence of infantile diarrhea would have been mostly determined by the prevalence of breastfeeding and the quality of the market milk, while for the case fatality rate, there would have been little difference between urban and rural areas, since no efficacious medical treatments for infantile diarrhea were yet available to either area.
For the late 19th through the early 20th centuries, there are no nationwide data on breastfeeding trends by urban and rural areas. However, mothers appear to have breastfed their infants less in urban areas than in rural areas. The U.S. Children's Bureau surveyed breastfeeding practices in selected areas of the country. In rural Kansas, a majority (92%) of mothers breastfed their infants in their first month, and over 61% continued to breastfed at six months (U.S. Children's Bureau 1917b). In a rural county in Kentucky, 86% of infants were breastfed to the sixth month of life and, among these, only 10% were weaned before one year of age (U.S. Children's Bureau 1922). In contrast, in an industrial city, Manchester, New Hampshire, about 82% of mothers breastfed their infants in their first month, and only 33% continued to do so at six months (U.S. Children's Bureau 1917a).
Circumstantial evidence indicates that urban milk was worse than rural milk. As urbanization progressed, sources of milk supply were increasingly located farther away from city centers. In the smaller cities and towns, the milk was transported by wagons, but as cities grew, more of the milk came by steamboat and railway (MacNutt 1917). In 1903, the Department of Agriculture surveyed the milk supply of 200 cities and towns (Alvord and Pearson 1903). In New York City, with a population of over 3 million, milk was sold from about 12,000 stores and 4,000 wagons. Milk was brought in from 3,500 dairy farms, mostly by rail. The longest shipment was 350 miles. Lansing, Michigan, then had a population of 16,485. Milk there was sold from three stores and 22 wagons. Milk was brought into Lansing on wagons from 45 dairy farms, from which the longest haul was four miles. In large cities, the milk usually arrived by train about midnight [End Page 595] and was picked up by vendors in the morning. It was usually delivered the following day, and was therefore at least 24 to 36 hours old when it came to the consumer (Parker 1917). The longer the distance or the time elapsed, the greater the likelihood that the milk would be contaminated.
Thus, higher diarrheal infant mortality in urban areas might have been due both to the lower prevalence rates of breastfeeding and to the use of poorer-quality market milk. The disparity in diarrheal infant mortality between urban and rural areas narrowed rapidly during the first two decades of the 20th century. By the 1930s, this disparity reversed, and cities had lower diarrheal mortality than rural areas. During this period, it was large cities that led the way by cleaning the market milk supply (Alvord and Pearson 1903; Fuchs and Frank 1938).
Impact of Diarrheal Diseases on Infant Mortality from Other Causes
Diarrhea had a major impact on infant mortality as an immediate culprit, but it also had other consequences that might have led to infant death. In their widely used and influential pediatric textbook, Holt and Howland (1923) stated that malnutrition was caused largely by ignorance, neglect of the common rules of hygiene, or bad feeding. They further stated that the most striking thing about children suffering from severe grades of malnutrition was their vulnerability to infection. In recent decades, it has been clearly demonstrated that there is a synergism between malnutrition, infection, and mortality (Keusch 2001).
A large number of infants who had survived diarrheal diseases would have suffered from lingering malnutrition that predisposed them to various infections, which in turn, once contracted, would have caused more serious illness or death. In Boston in 1911, the risk of death from non-diarrheal diseases among bottle-fed infants was five times greater than that of breastfed infants (Davis 1913). Over the period from 1924 to 1929, Grulee and his associates studied 20,061 infants in the Infant Welfare Society of Chicago (Grulee, Sanford, and Herron 1934). Of the total, 48.5% were entirely breastfed, 43% were partially breastfed, and 8.5% were artificially fed. Compared to exclusively breastfed infants, partially breastfed infants had a mortality rate seven times higher, and exclusively artificially fed infants had a mortality 60 times higher from infectious diseases, including respiratory infections.
Davis's study for Boston in 1911 showed a rate of overall infant mortality—between two weeks to 12 months of age—of 27 per 1,000 for the breastfed infants, and of 217 per 1,000 for the artificially fed infants. The Children's Bureau study of eight cities in the 1910s reported postneonatal infant mortality (excluding deaths from early infancy and from congenital malformations) of 42 per 1,000 for the breastfed, and of 159 per 1,000 for the artificially fed. Artificial feeding was associated with 54.5% all postneonatal infant deaths from both diarrheal [End Page 596] and non-diarrheal diseases in Boston in 1911, and 42.2% of all such infant deaths in eight cities in the United States between 1911 and 1915.
Diarrheal Diseases and the Decline in Infant Mortality
Before 1915, there was no nationwide birth registration in the United States. Examining the period between 1916 and 1950, when disease-specific infant mortality was based on birth and death registrations, infant mortality declined from 96.7 to 29.2 per 1,000. Among all causes, diarrheal mortality declined most remarkably, by 95.5%, from 23.4 per 1,000 live births in 1916 to 1.1 per 1,000 live births in 1950. Reduction in mortality from diarrhea alone was responsible for one-third of the total decline in infant mortality in this period. Out of the total decline in postneonatal mortality, diarrhea alone was responsible for 43% of this decline. All other infectious diseases combined were responsible for 38% of this decline. Thus, these two groups of diseases were responsible for 81% of the total decline in postneonatal infant mortality.
Again, this decline in diarrheal mortality had to come from either a reduction in the prevalence of diarrhea or a reduction in case fatality through medical intervention, or both. Major medical therapies for diarrhea, such as fluid and electrolyte replacement and correction of metabolic acidosis, were not widely practiced until the third or fourth decade of the 20th century (Kleinman, Barness, and Finberg 2003). Thus, it is likely that the decline in diarrheal mortality during this period was mainly from a decrease in the prevalence of diarrhea, rather than from medical intervention.
In turn, the prevalence of infantile diarrhea would have been mainly determined by two factors, the prevalence of breastfeeding and the quality of market milk. The Children's Bureau study revealed that the breastfeeding rate was 71.4% for the first three months, and 51.7% for the first six months. Hirschman and Butler (1981) examined breastfeeding practices in the United States, using cohorts of mothers born in succeeding decades of the 20th century. Among mothers born between 1911 and 1915, 55% of them continued their breastfeeding for the first three months, and 43% continued until six months of age. This rate declined among mothers born between 1921 and 1925: about 38% of them breastfed their infants for the first three months, and 27% for the first six months. Despite this decline in breastfeeding, diarrheal infant mortality continued to decrease, suggesting that the progressive decline in diarrheal mortality in the first half of the 20th century had to be mainly from the improvement in the quality of market milk.
During this period, infant mortality from infectious diseases declined as well, despite the absence of antibiotics and no widespread immunization for most of the common infectious diseases. Again, this decline in infant mortality from infectious [End Page 597] diseases might well have been due in part to the overall improvement in the nutritional status of an infant population with reduced prevalence of diarrhea.
At the turn of the 20th century, Newsholme (1899) regarded infant mortality as the most reliable index of the sanitary condition of a population. Newman (1906) stated that "the problem of infant mortality is not one of sanitation alone, or housing, or indeed of poverty as such, but is mainly a question of motherhood." Examining the decline in infant mortality in the United States in the first decade of the 20th century, Hibbs (1916) concluded: "The fundamental cause of the excessive rate of infant mortality in industrial communities is poverty, inadequate incomes, and low standards of living with their attendant evils" (p. 126). Thus, in the early part of the 20th century, infant mortality was regarded as a global socioeconomic problem.
Early in the 20th century, people did not have the scientific knowledge to link impure milk to infantile diarrhea, but many believed that cleaning market milk would reduce infant deaths from diarrhea. Nonetheless, many remained primarily concerned with unfavorable socioeconomic and generally poor sanitary conditions as factors leading to infant mortality. They failed to fully recognize the magnitude and the direct nature of the relationship between the quality of the market milk supply and overall infant mortality. Beaver (1973) was the first to suggest that the decline in infant mortality in the first part of 20th century derived primarily from cleaning the market milk supply. He observed that the infant mortality rate in England and Wales did not change much throughout the 19th century, despite economic growth, improving environmental health, and developing medical science. He stressed that the onset of the decline in infant mortality at the turn of the 20th century coincided with the beginning of movements for a safe milk supply and that it continued to decline with further improvement in the market milk supply. McKeown (1976) attributed the rapid fall in mortality in Britain to general economic growth, changes in the standard of living, and especially improved nutrition, rather than to any specific interventions to reduce mortality levels. He concurred with Beaver, however, that safe market milk was one of the major factors for the decline in infant mortality in the first part of the 20th century. Preston and Haines (1991) attributed the decline in childhood mortality to the diffusion of new technologies, public health interventions, and new parental knowledge in care of their children. Condran and her associates studied the impact of specific public health interventions on the decline in mortality, often by age and disease, in various regional populations in the United States (Cheney 1984; Condran and Cheney 1982; Condran and Lentzner 2004; Condran, Williams, and Cheney 1984). They argued that public health interventions such as water filtration, sanitation, and a clean milk supply played a major role in the shift in mortality in the late 19th and earlier 20th centuries, [End Page 598] and that these factors were far more important than income or nutrition. However, in their studies the decline in infant mortality was not attributed to a specific public health intervention.
To relate the decline in infant mortality to improvement in the quality of market milk would require concurrent longitudinal population data on the quality of market milk, the prevalence of infantile diarrhea, and diarrheal mortality, as well as concomitant information on socioeconomic and environmental factors associated with infant mortality. Lacking such information, this report has systematically reviewed and, when possible, reanalyzed historic cross-sectional data of different populations over different times in the late 19th and early 20th centuries. I conclude that, in late 19th and early 20th centuries, infantile diarrhea was a dominant cause of infant mortality in the United States. Diarrhea seemed to be directly and indirectly responsible for about half of all infant deaths, and it was one of the major causes of malnutrition in infancy. Artificial feeding was associated with an increased infant mortality from diarrhea, and this relationship was mediated through impure milk: low-income families, which more often consumed cheap, poor-quality market milk, had disproportionately high diarrheal infant mortality. Infant mortality, particularly diarrheal mortality, continued to decline without interruption in the first part of the 20th century, despite a progressive decline in breastfeeding and little advance in medical treatment for infantile diarrhea; the decline in infant mortality largely correlates with the cleaning of the market milk supply between 1840 and 1940. Concurrent improvement in living environments, such as water purification, other sanitary measures, and improved housing, had little, if any, effect on infant mortality due to diarrhea.
From these observations, this report concludes that cleaning the market milk supply was the single most important contributing factor to the decline in infant mortality in the United States during the first part of the 20th century, and that it played a far more important role than rises in income, other sanitary measures, or any medical interventions.