Preventability of neonatal cold injury and its contribution to neonatal mortality.

When the body temperature of a small neonate falls below 35 degrees C, lassitude can be noted; severe derangements of cardiovascular, renal, hepatic, immunological, and hematological systems may also occur depending in part on the duration and severity of hypothermia. Diagnosis requires a low-reading thermometer, lacking which the diagnosis can be suspected, but most often is missed. Fatal cases of diagnosed cold injury commonly have terminal pneumonia or sepsis. Prevention involves identification and home visits to high-risk infants; intensive care of those with the diagnosis at Soroka Hospital Medical Center has reduced the case-fatality rate from 30% in 1971 to 3% in 1988-1989. During the same period in our region, the proportion of neonatal deaths occurring in winter months of December, January, and February has dropped from 55 to 27%. The expected proportion is 25%. We hypothesize that excess neonatal mortality during winter months, especially due to pneumonia and sepsis or sudden infant death syndrome (SIDS) is an indicator of missed cold injury syndrome. A preliminary evaluation was made form U.S. data by state, provided by the National Center for Health Statistics, which records no fatalities from cold injury during 1986. Contrasted with this are 26 cold injury deaths in Israel for 1977-1980. In the U.S., though, excess winter neonatal deaths in 1986 from SIDS, pneumonia, and sepsis are reported.(ABSTRACT TRUNCATED AT 250 WORDS)


Introduction
Neonatal hypothermia was first designated "cold injury" by Nassau in 1948 (), but even earlier, infants with the same type ofdamage to the skin were designated as sclerema neonatorum by Underwood in 1811, who noted thatbabies presenting with induration ofsubcutaneous tissue had a low body temperature and a poor prognosis (2). Mann and Elliot in 1957 wrote (3): Commonly the disorder is miiagosed as hemorrhagic pneumonia or sclerema, the predominant role of exposure to cold being overlooked. Cases generally arise aferhome onfinements, especialy insevely cold weather.... Presenting features may include increasing apathy, food refusal, coldness to touch and oliguria: but the infant does not look ill.
The most constant clinical findings are hypothemia, oedema ofthe extremities, pumledt nasal skin aytm. heltreatment advised is slow rewarming, liberal glucose administration by intragastric drip, and antibiotic cover. Subsequent studies have been reported from Scotland, and seemingly paradoxically from a variety ofsubtropical countries, such as Israel, Ethiopia, India, and Iraq (4)(5)(6)(7)(8)(9). The seeming *Epidemiology and Health Services Evaluation Unit, Ben Gurion University of the Negev, P.Q. Box 653, Beer Sheva, Israel. tRegional Office for the Negev of the Ministry of Health. tSocial Pediatic and Obstetric Researh Unit, University of Glascow. Address reprint requests toJ. R. Goldsmith, Epidemiology and Health Services Evaluation Unit, Ben Gurion University ofthe Negev, P.Q Box 653, Beer Sheva, Israel. paradox is related to the fact that in these subtropical countries, cold nights are exceptional and therefore supplemental heat in sleeping rooms is not usually needed for adults. In addition, in these countries, many mothers are poorly educated.
Only four cases have been reported from the U.S. in the periodical literature. In the U.S. Vital Statistics for 1986, there were no deaths for which ICD-9 no. 778.2, "cold injury syndrome" (CIS) was the principal cause ofdeath, although the condition was mentioned, but not considered the underlying cause of death on one certificate.
In the U.S. in 1987 there were 368 neonatal deaths attributed to sudden infant death syndrome (ICD-9, 798.0), whereas in Israel for 1977-1980 among all infant deaths, 26 were attributed to 778.2, while 99 were attributed to SIDS. Ifwe limit the comparison in Israel to neonatal deaths the ratio of CIS deaths to SIDS deaths would be greater. There are subtropical areas ofthe U.S. with mothers oflow educational level and cold snaps. CIS deaths must be occurring in the U.S., but are not being recognized. Unless CIS is recognized, it will not be prevented, and, as we will show, CIS is preventable. We will look at possible indices that suggest where and when to look.

Natural History of Cold Injury Syndrome
Several studies in Israel called attention to important features ofthe natul history ofCIS. Cohen (4) called attention to the importance ofcoagulation abnormalities. Dagan and Gorodischer (10) reported on the frequency of infection among hypothermic infants. Meanwhile, a debate was going on as to whether it was preferable to rapidly or slowly warm the infants.

Hypothesis
It must follow that if in the presence of recognized hypothermia (24-34°C), one or more severe distubances are found, ifthe hypothermia is not recognized, these severe abnormalities would still be present and without vigorous treatment, are likely to be fatal. Under these circumstances, pneumonia, sepsis, respiratory failure, or SIDS are the most likely diagnoses. This hypothesis will be examined under two circumstances. First, as CIS becomes more consistently recognized (and treated) in the Negev, we will look to see what happens to mortality from pneumonia, sepsis, respiratory failures and SIDS in the neonatal period. A parallel examination will be made for the residents of the Gaza strip. If indeed, winter excess deaths from these conditions drops as CIS in recognized and treated or prevented, it will suggest that some cases ofCIS may have been missed in the earlier period but not in the later period.
Second, we will examine the neonatal mortality data for U.S. by state for the year 1986, looking for winter excess in such diagnoses as discussed previously. In locations with such a winter excess, it is reasonable to suspect that missed CIS is the cause, and an attempt to identify it would be justified.

Observations on Cold Injury in the Negev (South) of Israel
The incidence ofcold injury among infants has been a matter of serious concern among public health nursing personnel in the Negev Regional Office of the Ministry of Health for nearly 15 years. The nurses are equipped with low-reading thermometers (Fig. 3) and as part of the routine postnatal care, make home visits, with special attention being given to babies with small birth weight during the periods of cold weather. When babies with a body temperature of 34°C or less are identified, they are sent to the Soroka Hospital Medical Center.
Dagan and Gorodischer (10) reported on 51 such infants who were less than 3 months old and admitted during the years of 1976 ffirough 1981. Twenty-eight were males and 23 females; 41 were admitted during the first month oflife. Forty-three ofthem were from rural areas and villages, and 8 were born at home rather than in the hospital. Thirty-four were at least partially breast fed at the time of admission. Twenty-seven of them had some evidence of infection.
All were within the first 10th percentile ofweight for age, and 11 had diarrhea and dehydration. The tendency to dehydration was evidenced by the high prevalence ofelevated hematocrit (34 greater than 16 g hemoglobin/dL). Six died, of which 5 had infection.
A further 56 hypothermic infants were observed in the Soroka Hospital by Sofer et al., covering the period from November 1, 1982, to October 31, 1985. Of these, 29 were Bedouin and 27 Jews, which represents a slight shifttward Jews compared to 29 Bedouin and 22 Jews studied by Dagan and Gorodischer (JO). Forty-six of the babies studied by Sofer et al. were seen during the first 30 days of life (11). In contrast to the mortality rate of hospitalized infants prior to 1974, which was 31%, only 3 ofthe infants treated by Sofer et al. succumbed (11).
During the two winters, 1987-1988 and 1988-1989, the Public   Ifone restricts the consideration to high-risk infants born during cold weather, the incidence may be as high as 3 %. Table 1 shows the criteria for diagnosis of CIS. Table 2 shows the risk factors for cold injury by host and environment.  Solid bars, cold injury; hatched bars, pneumonia and sepsis; dotted bars, neonatal deaths. Theproportionofcold injurydeaths inthe winterisexpected to be high, and the proportion and numbers are shown. The proportion attributed topneumonia andsepsisandtheproportionofall neonatal deathsoccurrin duringthesemonthshasdropped towardtheexpected value of25 % as theeffectiveness ofcase-findingandpreventionofcold injury has improved. l977-1980and 1983-1985. Figure 4 showstheproportionsofthese deahstatoccurduring thewintermonthsofDecember, January, and February. Figure5 shows themonthly incidenceofthesedata for two timeperiods for the population ofthe Negev, and Figure  6 shows similar data for the Gaza Strip. Data for the U.S., 1986 and 1987 Table 4 shows the 20 states of the U.S. with elevated neonatal mortality rates in 1986, along with the numbers of such deaths, as well as the numbers in 1987 attributed to SIDS (ICD 798.0), influenza and pneumonia (ICD 480-487), and to other infections specific to the perinatal period (ICD 771.8). This latter category is elsewhere identified as "sepsis." The District of Columbia had the highest neonatal death rate and is included although not strictly a state.
The expected numbers of winter deaths can be obtained by multiplying the denominator data by 90/365 = 0.2466, the proportion ofdays in the year that occurs in the three winter months of Liecember, January, and February. The 20 states with neonatal mortality above the U.S. average shown in Tables 4 and 5 account for 18 ofthe 28 excess neonatal deaths, indicating that total neonatal mortality is not a very strong predictor ofwinter excess neonatal mortality from these causes. Of course, statewide data can obscure more striking problems in a low-income, badly housed minority group. The data show more striking effects for the first two categories, SIDS and influenza and pneumonia, than for the category designated as "sepsis." For New York City, to take an example, the data for SIDS are 4/9, for influenza and pneumonia 2/7, and for sepsis 7/32. For the SIDS and influenza and pneumonia, together, the winter ratio is 6/16 or 37.5%, whereas forthe state including New York City, it is 8/31 or 25.8%.
Overall, there are about 28 excess winter neonatal deaths from these three causes in the U.S. About half are due to SIDS and about equal numbers but proportionally more due to influenza and pneumonia than to sepsis.
The excess is greater in states ofthe southeastern region, where other risk factors for cold injury are likely to be high. A clinical trial ofthe use oflow-reading thermometers in such states would be justified.

Discussion
Most of cases of SIDS occur after the neonatal period, so it must follow that only a small proportion ofthe otherwise unexplained sudden infant deaths could be due to undiagnosed cold injury. According to our hypothesis, that proportion would be greatest in winter and in locations with high risk populations. Neonatal mortality data by state in the U.S. must include many deaths due to immaturity and congential abnormalities, as well as deaths from populations not at high risk. Despite this dilution, the data suggest that as much as 10% ofneonatal deaths attributed to SIDS or to influenza or pneumonia in certain states may be actually due to unrecognized cold injury. Since we have shown that deaths from cold injury, and to some extent, morbidity from cold injury, are preventable, the effort to detect and prevent these deaths seems worthwhile.