Malicious Use of Pharmaceuticals in Children

Article Outline

• Abstract
• Methods
  • Definitions
  • Data Analysis
• Results
• Discussion
• Acknowledgment
• Table III. 
• References
• Copyright

Objective

To describe malicious administration of pharmaceutical agents to children.

Study design

We performed a retrospective study of all pharmaceutical exposures involving children <7 years old reported to the US National Poison Data System from 2000 to 2008 for which the reason for exposure was coded as “malicious.”

Results

A total of 1439 cases met inclusion criteria. The mean number of cases per year was 160 (range, 124 to 189) that showed an increase over time. The median (IQR) age was 2 (1.5) years. Outcome data were available for 1244 (86.4%) patients. Of these exposures, 172 resulted in moderate or major outcomes or death. 9.7% of cases involved >1 exposed substance. The most common reported major pharmaceutical categories were analgesics, stimulants/street drugs, sedatives/hypnotics/antipsychotics, cough and cold preparations, and ethanol. In 51% of cases there was an exposure to at least one sedating agent. There were 18 (1.2%) deaths. Of these, 17 (94%) were exposed to sedating agents, including antihistamines (8 cases) and opioids (8 cases).

Conclusions

Malicious administration of pharmaceuticals should be considered an important form of child abuse.

AAPCC, American Association of Poison Control Centers, NPDS, National Poison Data System

Child abuse is a significant cause of preventable morbidity and mortality in children. The Department of Health and Human Services and the American Academy of Pediatrics list 4 primary forms of child maltreatment: neglect, physical abuse, sexual abuse, and emotional abuse. In 2006, more than 3.5 million children were investigated for child maltreatment, with 905 000 victims identified; 144 800 of these were victims of physical abuse.¹ In addition, homicide is one of the leading causes of injury related deaths in infants in the United States.² The true incidence of child abuse is also likely much higher because of underreporting and/or underdetection.³, ⁴

The nontherapeutic use of pharmaceuticals does not cleanly fit into the traditional classification of child maltreatment. Cases of homicide with pharmaceuticals and nonpharmaceuticals have been sporadically reported.⁵, ⁶ A review of pediatric autopsies showed that 6% of drug related deaths were homicides.⁷ Munchausen by proxy syndrome via poisoning has been well described.⁸, ⁹, ¹⁰ A recent review of pediatric fatalities associated with non-prescription cough and cold medicine suggested that malicious intent was a factor in some of these deaths.¹¹ In a separate report of 10 deaths related to cough and cold medicines, two were believed to be the result of child abuse.¹² Drug-facilitated sexual abuse in children has also been described.¹³

Based on these reports, it seems reasonable that nontherapeutic administration of pharmaceuticals to children may be a form of child maltreatment. The malicious use of pharmaceuticals has not been systematically characterized and primarily exists in the literature as case reports. The most recent American Academy of Pediatric guideline on the evaluation of suspected physical abuse or sexual abuse does not even mention investigating the malicious use of pharmaceuticals as an adjunct to the initial evaluation.³, ¹⁴ We believe that the malicious use of pharmaceuticals may be an under-recognized form and/or component of child maltreatment. The purpose of this study was to describe patterns of malicious administration of pharmaceutical to children reported to US poison centers in children <7 years old.

Back to Article Outline

Methods 

This is a retrospective database study characterizing all pharmaceutical exposures in children <7 years old reported to the National Poison Data System (NPDS) and coded as “malicious.” The University of Colorado Institutional Review Board approved this study.

Cases were identified from the NPDS (a database of all calls to US poison centers). A query was performed for the years 2000 to 2008 for all cases meeting the following criteria: the intent was coded as “malicious,” age was <7 years old, and at least one agent was a pharmaceutical or ethanol. All nonpharmaceuticals except for ethanol were excluded. Ethanol was specifically included because it is widely available, is one of the most commonly abused substance in the US, and has been well documented to facilitate malicious activity (sexual abuse, for example).¹⁵ The data collected were in accordance with American Association of Poison Control Centers (AAPCC) guidelines, and each case summary in NPDS represents an electronic summary of the original case data using standard codes. All data are collected, coded, and entered by trained professionals (specialists in poison information). Data collected for each case included age, sex, month and year of the exposure, the exposed substance or substances, amount ingested, disposition of patient, and poison center outcome designation. The specialists receiving the phone calls also coded the intent of each ingestion.

Definitions 

Whether an agent was a pharmaceutical or not was determined strictly by its AAPCC classification. Briefly, every substance is classified as either a pharmaceutical or a nonpharmaceutical and has a major category, minor category, and generic category designation. For example, methadone is classified under the major category of “analgesic,” minor category of “opioids,” and generic category of “methadone.” Not all individual pharmaceuticals have their own generic category. For example, lorazepam and diazepam would fall under the generic category of “benzodiazepines.” The 2008 AAPCC annual report provides a complete listing of all pharmaceutical categories.¹⁶ As defined by the AAPCC, “malicious intent” captures “patients who are victims of another person's intent to harm them.” Poison center outcome designations are coded according to the AAPCC definitions. They are designated and defined as follows: (1) no effect (the patient did not develop any signs or symptoms as a result of the exposure); (2) minimal clinical effects possible (no follow-up calls were made to determine the patient's outcome because the exposure was likely to result in only minimal toxicity; (3) minor (the patient had signs or symptoms as a result of the exposure); (4) moderate (the patient exhibited signs or symptoms as a result of the exposure that were more pronounced or more systemic in nature than minor symptoms but were not life-threatening and there was no residual disability); (5) major (the patient had signs or symptoms as a result of the exposure that were life threatening or resulted in significant residual disability); (6) unable to follow: potentially toxic effect (the patient was lost to follow-up, refused follow-up, or was not followed, but the exposure was significant and may have resulted in a moderate, major, or fatal outcome); and (7) unrelated: probably not responsible (the exposure was probably not responsible for the effect).

The investigator (a medical toxicologist) determined whether a pharmaceutical was sedating and these included opioids, antipsychotics, antidepressants, antihistamines, ethanol, anticonvulsants, imidazolines, muscle relaxants, and benzodiazepines.

Data Analysis 

Data were analyzed with SAS, version 9.1.3 (SAS, Inc., Cary, North Carolina). Cases were analyzed with descriptive statistics. Linear regression was used to determine trend over time. From the available data, three potential risk factors for worse outcomes were identified which were (1) exposure to a sedating agent; (2) exposure to multiple agents; and (3) age ≤2. Odds ratios for death and major outcomes for these risk factors were determined using standard χ² analysis. Any factor that was significant (P < .05) was placed in a logistic regression model to calculate adjusted odds ratios. Death/major outcomes were compared with all other outcomes except cases that were coded “unable to be followed, potentially toxic effect,” which were removed for any outcome analysis. These were removed from outcome analysis because they may have potentially resulted in moderate or major outcomes or death. If the pharmaceutical exposed was unknown, then those cases were removed for analysis relating to exposure to a sedating agent.

Last, poison center fatality abstracts for any deaths in this study cohort were examined. These abstracts are summaries of the poison center narrative prepared at the initial time of the exposure. They document any circumstances relayed to the poison center surrounding the exposure and subsequent course. These were examined to see if the circumstances were consistent with a malicious exposure.

Back to Article Outline

Results 

Of approximately 21.4 million exposures reported to NPDS during the study period, 1439 (0.007%) cases met inclusion criteria comprising 1634 total exposures. The mean number of cases per year was 160 (range, 124 to 189). Cases increased over time in a linear model (P = .006). Boys were involved in 56.6% of the cases. The median (IQR) age was 2 years (1.5). Table I shows the distribution of ages and outcomes. Outcome data were available for 1244 (86.4%) patients. Of these cases, 13.8% resulted in moderate or major outcomes or death.

Table I. Distribution of ages and outcomes

Age (years)
0-<1	311 (21.6%)
1-<2	220 (15.3%)
2-<3	202 (14.0%)
3-<4	157 (10.9%)
4-<5	165 (11.5%)
5-<6	176 (13.3%)
6-<7	192 (13.3%)
Unknown	16 (1.1%)
Outcomes
Death	18 (1.2%)
Major	32 (2.2%)
Moderate	122 (8.5%)
Minor	290 (20.1%)
Minimal effects	321 (22.3%)
No effect	323 (22.4%)
Unrelated	69 (4.8%)
Potentially toxic	195 (13.5%)

Table II shows the distribution of the 11 most commonly reported major category classes with the most common generic categories within each major class. These 11 comprised 82.3% of all the exposures mentioned. There were 16 other major category classes mentioned with none compromising more than 3%. The most common generic category mentions were unknown (n = 155), ethanol (n = 111), laxatives (n = 67), benzodiazepines (n = 62), acetaminophen and dextromethorphan combinations with decongestant and/or antihistamine without phenylpropanolamine (n = 62), and diphenhydramine (n = 60). In 9.7% of the cases, there was an exposure to more than one pharmaceutical. In 51.1% of cases, there was an exposure to at least one sedating agent.

Table II. Most commonly reported major and generic drug categories

Analgesics	176
Acetaminophen alone	50 (28.4%)
Ibuprofen	30 (17.0%)
Acetaminophen in combination with other drug	16 (9.1%)
Stimulants/street drugs	173
Marijuana	42 (24.3%)
Cocaine	30 (17.3%)
Methamphetamine	27 (15.6%)
Sedatives/hypnotics/antipsychotics	158
Bzenodiazepines	62 (39.2%)
Atypical antipsychotics	52 (25.0%)
Other type of sedative/hypnotic/antipsychotic	19 (12.0%)
Cold and cough preparations	155
Acetaminophen and dextromethorphan combinations with decongestant and/or antihistamine without phenylpropanolamine	62 (40%)
Antihistamine and/or decongestant with codeine without phenylpropanolamine	27 (17.4%)
Antihistamine and/or decongestant without phenylpropanolamine and opioid	25 (16.1%)
Unknown drug	155
Ethanol (beverage)	111
Topical preparations	106
Other type of topical antiseptic	24 (22.6%)
Diaper care and rash products	13 (12.3%)
Methyl salicylate	13 (12.3%)
Hydrogen peroxide	11 (10.4%)
Gastrointestinal preparations	89
Laxatives	67 (75.3%)
Other type of gastrointestinal preparation	8 (9.0%)
Antihistamines	85
Diphenhydramine alone	60 (70.6%)
Other antihistamines alone	23 (27.1%)
Antidepressants	75
Selective serotonin reuptake inhibitor	33 (44.0%)
Trazadone	17 (22.7%)
Amitriptyline	12 (16.0%)
Cardiovascular	62
Clonidine	38 (61.3%)
β-blockers	6 (9.7%)
Calcium antagonists	6 (9.7%)

Eighteen children (1.2%) died (Table III; available at www.jpeds.com) . The mean age of decedent children was 1.6 years. Seventeen (94%) of the cases involved an exposure to at least one sedating agent, including antihistamines (8 cases) and opioids (8 cases). Fatality abstracts were available for review in 17 of 18 cases. Eleven of the cases documented situations in which malicious intent was likely including: four ruled homicide by a coroner, three with legal action against the mother, two with highly suspicious circumstances, one with the presence of an illicit agent (cocaine), and one confession. The other six cases did not have enough information to determine intent based solely on the abstract. Confirmation of pharmaceutical exposure was noted with either quantitative or qualitative measures in 15 of 17 cases. Only case 17 commented on the presence or absence of traumatic injuries, which were not present in the case.

Exposure to multiple substances, exposure to a sedating agent, and age ≤2 years were significantly associated with death or major outcomes in univariate modeling. These three factors were placed in a logistic model and the adjusted odds ratios of all three factors remained significant (Table IV).

Table IV. Odds ratios for risk factors for major outcome and death

Back to Article Outline

Discussion 

On average, 160 cases per year were reported to US poison centers during the study period, including an average of two deaths per year. Cases reported to poison centers increased over time. As a practical matter, only exposures to pharmaceuticals were included in this study. This would only underestimate the true incidence of child maltreatment by poisoning as cases with nonpharmaceuticals such as caustic agents,¹⁷ rodenticides,¹⁸ and ethylene glycol¹⁹ have been reported. Although Kempe et al²⁰ mentioned the aspect of intentional poisoning in first describing the battered child in 1962, there has not subsequently been a systematic study of the topic. Previous literature is generally limited to sporadic case reports or series. Poisoning may also be occurring in conjunction with other forms of abuse. In a published abstract of pediatric adverse events detected in 2008 from malicious use of cough/cold medicine, three of six fatal cases had signs of physical abuse.²¹ In a 1982 review of intentional poisonings, seven of 41 cases were associated with battering.²² If even a fraction of the nearly 150 000 annual cases of physical abuse also involve an element of nontherapeutic pharmaceutical administration, then we are missing thousands of cases each year. Examination of Centers of Disease Control mortality statistics show that approximately 14 cases of homicide by poisoning in children <5 per year occurred from 2000 to 2006.²³ Hence, there is at least a sevenfold under-reporting of fatalities to NPDS. Because it is probable that deaths are more likely to be investigated and thus found to be malicious in nature, the underestimation of nonfatal cases is probably even higher. This study provides a starting point for estimating the true incidence of this potentially under-recognized problem.

In this study, 13.8% of the cases in which the outcome was known resulted in an outcome that was moderate or worse. As a comparison, in 2007 only 0.9% of all exposures in children <6 years of age reported to NPDS resulted in moderate or worse outcome, which is more than a 15-fold increase. In our study cohort, 1.2% of cases resulted in a fatality compared with 0.004% of all exposures in children <6 years of age reported to NPDS in 2007, which is a 300-fold difference. These exposures are resulting in more severe outcomes than general pediatric poison exposures as would likely be expected with malicious poisoning. A reporting bias may exist in this case because again it is likely that deaths will be investigated more closely than nondeaths and perhaps nonfatal cases of malicious administration of pharmaceuticals did not get coded as such. This would increase the percentage of fatalities in our study.

A standard urine drugs of abuse screen may have only been positive in three of the fatalities in our study cohort (morphine, cocaine, and alprazolam) and perhaps six more if the drug screen included methadone (n = 4) and tricyclic antidepressants (n = 2) that may be included in some facilities' standard urine drug screen. Comprehensive drug screening/testing probably would have identified every case. Early recognition of child abuse is important as it may prevent more serious sequelae as one study showed that some victims of physical abuse are seen multiple times before recognition of their abuse and fatalities may have been prevented.²⁴ Comprehensive drug screening should perhaps be considered in the initial evaluation of a child suspected to be the victim of maltreatment. Of course, comprehensive drug screening has limitations. It is usually performed with either thin layer chromatography (qualitative) or liquid or gas chromatography/mass spectrometry (quantitative). These methodologies probably are limited to university associated labs, national reference labs, or government forensic labs and may take a number of days to be resulted. The screen is also limited by the reference library used for comparison to unknowns and may not cover uncommon or unusual agents. In addition, substances may only be detected in the urine or blood within a limited time after exposure (usually only a few days, depending on the agent) so a negative does not rule out exposure. Last, drug screening will not distinguish therapeutic use of an agent such as diphenhydramine although quantitative testing may reveal levels above expected from therapeutic dosing.

The motives in the malicious use of pharmaceuticals are purely speculative and probably are variable. Infant crying has been demonstrated to be an important precipitant to shaken baby syndrome.²⁵, ²⁶ It is not hard to imagine that a caregiver may choose to sedate a crying baby to get them to stop. Alternatively, a physically abused baby is likely to be more irritable than usual and the perpetrator may wish to sedate them after an inflicted injury. Perhaps a caregiver may be administering a sedating agent simply as a respite from the responsibilities of childcare. Even though the caregiver probably is not strictly intending to harm the child, this still constitutes maltreatment. In these instances, the child is likely to come to the attention of a physician (and a poison center) only if an adverse event has occurred or another party becomes aware of the situation. In this series, 51.1% of cases involved exposure to at least to one sedating drug and 17 of the 18 deaths involved at least one sedating drug. Use of a sedating agent was also 3.6 times more likely to be associated with a death or a major outcome. In the study by Dart et al,¹¹ sedation appeared to be the intent of many nontherapeutic uses of over-the-counter cough and cold medication in children. Other reasons for nontherapeutic administration of pharmaceuticals may include homicide,⁵ punishment,²⁷ Munchausen by proxy,¹⁰ or amusement.²⁸

Limitations of the study are related to its retrospective nature. There may have been miscoding. Coding by the specialists is currently done with drop-down menus and thus the coding of “malicious” may have simply been a coding error. As the severity of the outcome increased, there probably would be more oversight in the coding especially with respect to fatalities. The exact circumstances of each case are not known. It is likely that not every case coded as malicious was indeed a case of child maltreatment. However, review of the fatality abstracts demonstrates unequivocally that NPDS captures cases of child maltreatment via malicious administration of pharmaceuticals. In addition, the trend for decreasing incidence as the child age increases in our study is similar to Department Health and Human Services statistics on known victims of child maltreatment. Also the fact that children ≤2 years of age were more likely to suffer a worse outcome in this study is also reflective of what is known about pediatric victims of physical abuse.²⁹ Of course, children ≤2 years of age in general are more likely to have poorer outcome in unintentional ingestions as well simply because their lower body weight means a lower exposure is required for symptoms. Whether the age distribution in our study reflects a similar pattern to children who are victims of other forms of maltreatment or is a reflection of pediatric poisoning in general is not clear at this point. Poison center data also rely on voluntary and self-reporting of cases but are generally believed to be an underestimation of the true incidence of drug and poison exposures.³⁰, ³¹ In the case of nontherapeutic administration of pharmaceuticals, this is particularly likely because the perpetrators are not likely to fully disclose their actions and many cases are most likely not recognized as such. NPDS also first began accepting “indirect” death reports in the first year of the study period. This would again over represent deaths because NPDS does not incorporate “indirect” reports that do not result in death. Finally, other than the fatalities, data confirming exposures with drug levels were not available.

Back to Article Outline

The author thanks Jamie Kokko, MPH, for statistical assistance (employee of the Rocky Mountain Poison and Drug Center).

Back to Article Outline

Table III. 

Summary of fatalities

Back to Article Outline

References 

1. Legano L, McHugh MT, Palusci VJ. Child abuse and neglect. Curr Probl Pediatr Adolesc Health Care. 2009;39(31):e1–26
   • View In Article
2. Tomashek KM, Hsia J, Iyasu S. Trends in postneonatal mortality attributable to injury, United States, 1988-1998. Pediatrics. 2003;111:1219–1225
   • View In Article
3. Kellogg ND. Evaluation of suspected child physical abuse. Pediatrics. 2007;119:1232–1241
   • View In Article
   • CrossRef
4. Herman-Giddens ME, Brown G, Verbiest S, Carlson PJ, Hooten EG, Howel E, et al. Underascertainment of child abuse mortality in the United States. JAMA. 1999;282:463–467
   • View In Article
   • MEDLINE
   • CrossRef
5. Campbell TA, Collins KA. Pediatric toxicologic deaths: a 10-year retrospective study. Am J Forensic Med Pathol. 2001;22:184–187
   • View In Article
   • MEDLINE
   • CrossRef
6. Sauvageau A, Racette S. Child and adolescent victims in forensic autopsy: a 5-year retrospective study. J Forensic Sci. 2008;53:699–702
   • View In Article
   • CrossRef
7. Naso C, Jenkins AJ, Younger D. A study of drug detection in a postmortem pediatric population. J Forensic Sci. 2008;53:483–490
   • View In Article
   • CrossRef
8. Holstege CP, Dobmeier SG. Criminal poisoning: Munchausen by proxy. Clin Lab Med. 2006;26:243–253
   • View In Article
   • Full Text
   • Full-Text PDF (145 KB)
   • CrossRef
9. Giurgea I, Ulinski T, Touati G, Sempoux C, Mochel F, Brunelle F, et al. Factitious hyperinsulinism leading to pancreatectomy: severe forms of Munchausen syndrome by proxy. Pediatrics. 2005;116:e145–e148
   • View In Article
   • CrossRef
10. Meadow R. Munchausen syndrome by proxy. Arch Dis Child. 1982;57:92–98
    • View In Article
    • CrossRef
11. Dart RC, Paul IM, Bond GR, Winston DC, Manoguerra AS, Palmer RB, et al. Pediatric fatalities associated with over the counter (nonprescription) cough and cold medications. Ann Emerg Med. 2009;53:411–417
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (410 KB)
    • CrossRef
12. Marinetti L, Lehman L, Casto B, Harshbarger K, Kubiczek P, Davis J. Over-the-counter cold medications-postmortem findings in infants and the relationship to cause of death. J Anal Toxicol. 2005;29:738–743
    • View In Article
    • MEDLINE
13. Spiller HA, Rogers J, Sawyer TS. Drug facilitated sexual assault using an over-the-counter ocular solution containing tetrahydrozoline (Visine). Leg Med (Tokyo). 2007;9:192–195
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (84 KB)
    • CrossRef
14. Guidelines for the evaluation of sexual abuse of children: subject review. American Academy of Pediatrics Committee on Child Abuse and Neglect. Pediatrics. 1999;103:186–191
    • View In Article
15. McCauley JL, Conoscenti LM, Ruggiero KJ, Resnick HS, Saunders BE, Kilpatrick DG. Prevalence and correlates of drug/alcohol-facilitated and incapacitated sexual assault in a nationally representative sample of adolescent girls. J Clin Child Adolesc Psychol. 2009;38:295–300
    • View In Article
    • CrossRef
16. Bronstein AC, Spyker DA, Cantilena LR, Green JL, Rumack BH, Giffin SL. 2008 Annual Report of the American Association of Poison Control Centers' National Poison Data System (NPDS): 26th Annual Report. Clin Toxicol (Phila). 2009;47:911–1084
    • View In Article
    • CrossRef
17. Gotschlich T, Beltran RS. Poisoning of a 21-month-old child by a baby-sitter. Clin Pediatr (Phila). 1995;34:52–53
    • View In Article
    • MEDLINE
    • CrossRef
18. Babcock J, Hartman K, Pedersen A, Murphy M, Alving B. Rodenticide-induced coagulopathy in a young child: a case of Munchausen syndrome by proxy. Am J Pediatr Hematol Oncol. 1993;15:126–130
    • View In Article
    • MEDLINE
19. Saladino R, Shannon M. Accidental and intentional poisonings with ethylene glycol in infancy: diagnostic clues and management. Pediatr Emerg Care. 1991;7:93–96
    • View In Article
    • MEDLINE
    • CrossRef
20. Kempe CH, Silverman FN, Steele BF, Droegemueller W, Silver HK. The battered-child syndrome. JAMA. 1962;181:17–24
    • View In Article
    • MEDLINE
21. Yin S, Blair R, Green J, Dart R. Child Abuse with Cough/Cold Medications. Clin Toxicol. 2009;47:761
    • View In Article
22. Dine MS, McGovern ME. Intentional poisoning of children–an overlooked category of child abuse: report of seven cases and review of the literature. Pediatrics. 1982;70:32–35
    • View In Article
    • MEDLINE
23. CDC Wonder. Accessed January 1, 2010, at http://wonder.cdc.gov.
    • View In Article
24. Jenny C, Hymel KP, Ritzen A, Reinert SE, Hay TC. Analysis of missed cases of abusive head trauma. JAMA. 1999;281:621–626
    • View In Article
    • MEDLINE
    • CrossRef
25. Barr RG, Rivara FP, Barr M, Cummings P, Taylor J, Lengua LJ, et al. Effectiveness of educational materials designed to change knowledge and behaviors regarding crying and shaken-baby syndrome in mothers of newborns: a randomized, controlled trial. Pediatrics. 2009;123:972–980
    • View In Article
    • CrossRef
26. Reijneveld SA, van der Wal MF, Brugman E, Sing RA, Verloove-Vanhorick SP. Infant crying and abuse. Lancet. 2004;364:1340–1342
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (131 KB)
    • CrossRef
27. Baugh JR, Krug EF, Weir MR. Punishment by salt poisoning. South Med J. 1983;76:540–541
    • View In Article
    • MEDLINE
    • CrossRef
28. Schwartz RH, Peary P, Mistretta D. Intoxication of young children with marijuana: a form of amusement for 'pot'-smoking teenage girls. Am J Dis Child. 1986;140:326
    • View In Article
29. US Department of Health and Human Services AoC, Youth and Families. Child Maltreatment 2007, Washington DC.
    • View In Article
30. Blanc PD, Jones MR, Olson KR. Surveillance of poisoning and drug overdose through hospital discharge coding, poison control center reporting, and the Drug Abuse Warning Network. Am J Emerg Med. 1993;11:14–19
    • View In Article
    • MEDLINE
    • CrossRef
31. Blanc PD, Kearney TE, Olson KR. Underreporting of fatal cases to a regional poison control center. West J Med. 1995;162:505–509
    • View In Article
    • MEDLINE