Using Computer Decision Support to Increase Maternal Postpartum Tetanus, Diphtheria, and Acellular Pertussis Vaccination

Trick, William E. MD; Linn, Edward S. MD; Jones, Zina RN; Caquelin, Clay; Kee, Romina MD, MPH; Morita, Julia Y. MD

Obstetrics & Gynecology:
doi: 10.1097/AOG.0b013e3181e40a9f
Original Research

OBJECTIVE: To evaluate whether use of a computer-based clinical decision-support algorithm that used data stored in the electronic medical record increased administration of tetanus, diphtheria, and acellular pertussis (Tdap) vaccine to postpartum women.

METHODS: We performed a before and after cohort study of postpartum women at an urban public teaching hospital. We compared the frequency of Tdap vaccination during the preintervention (October 1, 2008–January 14, 2009) and postintervention (January 15–April 30, 2009) time periods. We intervened by automating electronic presentation of preselected orders to physicians who provided postpartum care. The order was displayed when physicians ordered iron supplementation or patient discharge to a woman who met certain criteria. We evaluated whether patient characteristics were associated with receipt of vaccine.

RESULTS: Tetanus, diphtheria, and acellular pertussis vaccination was more likely for postpartum women postintervention compared with preintervention (147 of 248 [59%] compared with zero of 183 [0%]; difference=59%; 95% confidence interval [CI] 53–65%). Among 248 women who delivered during the postintervention period, those who met pharmacologic criteria for decision support rule activation were vaccinated more often than those who did not meet criteria (146 of 232 [63%] compared with one of 16 [6%]; difference=57%; 95% CI 43–70%). Race and ethnicity and cesarean delivery were not associated with vaccine receipt; however, there was a lower likelihood of vaccination among older women (P=.05 by a trend test across age quartiles).

CONCLUSION: We implemented a computer-based clinical decision-support algorithm that dramatically increased Tdap vaccination of postpartum women. Deployment of our algorithm in hospitals that have clinical decision support systems should increase rates of this important postpartum preventive intervention.


In Brief

A computer-based decision-support algorithm using the electronic medical record dramatically increases postpartum maternal tetanus, diphtheria, and acellular pertussis vaccination.

Author Information

From the Collaborative Research Unit, Stroger Hospital of Cook County; the Department of Medicine, Rush Medical College; the Obstetrics and Gynecology Department, Stroger Hospital of Cook County; the Informatics Department, Stroger Hospital of Cook County; and the Immunization Program, Chicago Department of Public Health, Chicago, Illinois.

The Tdap vaccine was provided by the Chicago Department of Public Health.

Corresponding author: William E. Trick, MD, Collaborative Research Unit, Stroger Hospital of Cook County, 1900 W. Polk Street, Suite 1600, Chicago, IL 60612; e-mail:

Financial Disclosure The authors did not report any potential conflicts of interest.

Article Outline

Infection by Bordetella pertussis is common and has increased over the past several decades.1 These infections result in substantial morbidity for adults2 and can be catastrophic for infants.3,4 In the United States, most pertussis-related deaths occur among infants younger than 12 months of age,1 and most of these deaths occur before 6 months of age. Commonly, the source of infection is a close personal contact,5 with mothers being the most frequent identifiable source.6 Infants do not complete their tetanus, diphtheria, and acellular pertussis (Tdap) primary immunization series until 6 months of age; therefore, they remain susceptible to pertussis infection during early infancy. Because newborns have no protective immunity and because the Tdap booster vaccine is highly effective at preventing infection and subsequent transmission among adults and adolescents,7 the Advisory Committee of Immunization Practices recommends immediate postpartum vaccination for previously unvaccinated women.1 The American College of Obstetricians and Gynecologists8 supports these guidelines; however, vaccination strategies need to be developed.9 Despite an effective vaccine and support of Tdap vaccination by most obstetricians,10 at our hospital, previously unvaccinated postpartum women did not routinely receive Tdap vaccine.

One alternative to postpartum Tdap vaccination is to administer the vaccine to women of childbearing age during routine clinic visits; however, many young adult women do not receive routine medical care or do not have a primary care provider.11 For those who receive medical care during their pregnancy, Tdap vaccination during pregnancy has not been recommended as a result of a lack of data regarding vaccine safety1 and concerns that transplacental maternal antibody might blunt an infant's immune response to subsequent pertussis vaccinations.12,13 At our hospital, after an educational campaign directed at clinicians and nurses failed to increase postpartum Tdap vaccination, we implemented a computer-based decision support intervention modeled after an intervention we used to increase influenza vaccination among medical inpatients.14 We present the algorithm used for our computer-based decision support strategy and the results from our evaluation of the effectiveness of our intervention in increasing Tdap vaccination among postpartum women.

Back to Top | Article Outline


We performed a before and after study of postpartum women hospitalized at Stroger Hospital (formerly Cook County Hospital), a 464-bed, public teaching hospital located in the city of Chicago. We activated the decision-support algorithm on January 15, 2009; the preintervention period was during October 1, 2008, through January 14, 2009; the postintervention period was during January 15, 2009, through April 30, 2009. We included all postpartum women except those whose hospitalization overlapped the intervention initiation date. Our project was determined to be exempt from review by Stroger Hospital's Institutional Review Board.

Computer-based clinical decision support systems are applications incorporated into the hospital's information system; the system uses information in the electronic medical record to provide logic-based diagnostic or therapeutic guidance to clinicians. At our hospital, implementation of computer-based decision-support algorithms requires approval by the decision support committee, which is run by a physician informaticist in collaboration with the Information System Department. The clinical logic for the algorithm is discussed with a hospital informaticist who translates the logic into the separate rules that comprise the algorithm. Components of algorithm development include three separate domains as follows: 1) conditions (ie, inclusion and exclusion criteria); 2) one or more triggers, which activate the rule meant to change the physician's behavior (eg, ordering Tdap) if conditions are satisfied; and 3) the electronic method for realizing the change in physician behavior (eg, alert messages or preselected orders). Decisions for these components are informed through prior experience, literature review, and discussions with content experts (in this case, obstetricians and labor and delivery nurses).

The conditions for our algorithm were female 14–45 years of age (inclusive) and, to prevent vaccination of pregnant women before delivery, receipt of oxytocin during the current admission. In Illinois, postpartum women have the capacity to make medical decisions without parental consent regardless of age. Exclusion criteria were a prior electronic order for Tdap vaccine or tetanus–diphtheria (Td) vaccination within 2 years or phenytoin during the current admission. The trigger for display of the preselected Tdap order was iron supplementation (eg, ferrous sulfate or ferrous gluconate), because it routinely was prescribed after delivery. Because we recognized that not all women received iron and some were prescribed iron before oxytocin, on February 9, 2009, we included the discharge patient order as a second order display trigger. The logic for our algorithm is displayed in Figure 1.

If criteria were met, at the time a physician entered an order for iron supplementation, a dialogue box was displayed that contained a preselected order for Tdap vaccination that read “Tdap vaccination is recommended for postpartum mothers. An order for Tdap will be generated and sent to Pharmacy and Nursing unless you deselect the order below. If the patient has previously received this vaccine (Tdap), there is no need to repeat administration of the vaccine.” Typically, postpartum orders are entered by housestaff. Unless the physician actively deselected the order, the order was electronically sent to the pharmacy and nurse's electronic medication administration record. Physicians were educated about the decision support intervention during grand rounds and encouraged to comment on the algorithm's rules. Nursing leadership was instrumental in development of the algorithm and educating labor and delivery nurses about the intervention. Because it was not possible to search electronic records for prior vaccination at other healthcare facilities, physicians and nurses were instructed to inquire about a woman's vaccination history. For patients who could not recall their vaccine history, we recommended that they receive the Tdap vaccine. Women who reported a prior allergic reaction to Td or Tdap vaccine or a recent neurologic event, inclusive of a seizure, were not to receive the vaccine. We used the decision support system available through our hospital's information system (Cerner, Inc, Kansas City, MO).

We obtained data on participants' characteristics (eg, age and race and ethnicity group) and vaccine ordering and receipt through manual review of the electronic medical record and queries of data warehouses maintained by our hospital's information system and Department of Medicine.14 Nurses were instructed to document reasons for failure to administer the vaccine in the electronic medical record. Using the International Classification of Diseases, 9th Revision, Clinical Modification codes stored in the data warehouse (74.0, 74.1, 74.2, 74.4, V27.0, V27.1, V27.3, and V27.5), we identified a cohort of postpartum women. We joined pharmacy (ie, iron supplementation, oxytocin, phenytoin, Tdap orders and administration) and demographic data.

Our primary outcome was receipt of Tdap vaccination preintervention and postintervention. To estimate the sample size needed, we assumed a baseline vaccination rate of 5% (we expected a preintervention vaccination rate between 0% and 5%) and considered a 20% absolute increase in the rate of vaccination as clinically meaningful. After setting two-tailed α=.05 and power=.8, we needed to collect data on 49 women from each time period. Because we also wanted to evaluate patient characteristics associated with vaccination, we exceeded the requisite sample size and sampled all patients during an equivalent time period before and after we implemented the intervention.

We created indicator variables for characteristics that had multiple categories such as race and ethnicity classifications. To explore the association between age and Tdap administration, we generated graphic representations of the data through locally weighted scatterplot smoothing (lowess) methods.15 After observing a reduced likelihood of vaccination with increasing age, we further explored the association between receipt of Tdap and age by creating quartiles of age categories using the distribution of age in our data set. Then, the association between Tdap vaccination and the ordered age quartiles (ie, youngest to oldest) was evaluated using a nonparametric test for trend.16

We compared categorical variables using the chi square test. We compared continuous variables using Student t test; we report the two-tailed P values. We used the statistical software package Stata 10.1 (Stata Corporation, College Station, TX).

Back to Top | Article Outline


We evaluated 431 women, 183 from the preintervention and 248 from the postintervention periods. In comparing the two periods, women were similar in age, race and ethnicity, method of delivery, and length of stay; the rate of births per day was higher during the postintervention period (Table 1). Among 248 women admitted postintervention, most (238 [96%]) met the pharmacy condition for rule activation (ie, oxytocin order) and for most (93%) women who had an oxytocin order, there also was an order for iron supplementation (Fig. 2). Among the 222 women who met pharmacy conditions for the electronic Tdap vaccine order, for 180 (81%), the physician accepted the order (Fig. 2). There were 16 women who had an oxytocin order, but no pharmacy order for iron supplementation; only one (6%; 95% confidence interval [CI], 0–30%) of these women had a Tdap vaccine order placed (Fig. 2).

Overall, compared with the preintervention period, vaccine was ordered (73% compared with 0%; difference=73%; 95% CI 67–79%) and administered (59% compared with 0%; difference=59%; 95% CI 53–65%) dramatically more often during the postintervention period (Fig. 2). The increased rate of vaccination only was apparent for women who met pharmacy criteria for activation of the decision support rule (146 of 222 [66%] compared with 1 of 26 [4%]; difference=62%; 95% CI 52–72%); therefore, the intervention effect was unlikely to be the result of a time-dependent factor other than our intervention (Table 2). Compared with the first 4 weeks postintervention, vaccination coverage during the final 4 weeks was equivalent (difference=0%; P=.96). Physicians were equally likely to deselect the Tdap order during the first 4 weeks compared with the final 4 weeks (13 of 56 [23%] compared with eight of 41 [19%]; P=.66). Reasons for failure to vaccinate after a Tdap order had been placed (n=34) included patient refusal (n=30), prior vaccination (n=2), and reported allergy (n=1); one patient likely had been vaccinated but without appropriate nursing documentation of vaccination because the order comment included the lot number and expiration date.

When we evaluated patient-level factors associated with a Tdap vaccine order, we found that the likelihood of vaccination was lower among women in older age categories; however, this finding was of borderline statistical significance and may have been the result of chance (Table 2). We found no differences in the frequency of vaccination for vaginal compared with cesarean delivery or African-American women compared with women of other race or ethnicity classification (58% compared with 62%, P=.55).

Back to Top | Article Outline


Use of a computer-based clinical decision-support algorithm dramatically increased Tdap vaccination of women during their postpartum hospitalization. The increased frequency of vaccination was evident irrespective of race and ethnicity, method of delivery, and across all ages. Because we performed an effectiveness evaluation (ie, all postpartum patients were analyzed and our unfunded intervention only used existing personnel), our strategy should be exportable to hospitals that have computer-based decision support systems.

Before opting to use a computer-based intervention, we attempted to increase the rate of vaccination through educational interventions directed at both physicians and nurses; our relatively low-intensity educational efforts failed to increase Tdap vaccination rates. One option would have been to intensify our educational efforts to providers and expand the intervention to patients; however, because we previously had failed to increase inpatient influenza vaccination of general medicine patients through an intensive provider and patient educational program,14 we opted to develop a computer-based decision-support algorithm. We considered other system-directed options for increasing Tdap vaccination such as implementing a standing orders policy or a postpartum order set inclusive of Tdap vaccination. We did not choose a paper-based standing orders policies because this strategy was ineffective at increasing influenza vaccination at our hospital and during a separate multicenter evaluation.14,17 Also, prior implementation of a paper-based standing orders policy required substantial effort to educate nurses and inform potential vaccines.18 Similar to a standing orders policy, the intent of our computer-based decision-support algorithm was to facilitate vaccination of all postpartum women.

In a separate evaluation, Healy et al19 used standing orders to increase postpartum Tdap vaccination and achieved a vaccination rate that exceeded ours (72% compared with 59%); their success provides a model for institutions that have the resources to implement and sustain intensive provider, nurse, and patient education. In contrast, our computer-facilitated intervention allows for the maintenance of a vaccination intervention with minimal ongoing effort and cost. Interventions such as our Tdap decision support rule are one of the criteria specified by the Department of Health and Human Services in their interim final rule for health systems to demonstrate meaningful use of electronic health record technology.

Subsequent to our evaluation, we made changes to the algorithm to increase the sensitivity of detecting postpartum women. In particular, we added the following additional pharmacy triggers for display of the preselected order: receipt of a stool softener (ie, docusate or citric acid–sodium citrate), phytonadione, or acetaminophen–oxycodone. During our project, these triggers would have resulted in display of the Tdap order for an additional 13 of the 16 women who were not prescribed iron.

To minimize the likelihood of inappropriately vaccinating pregnant women, we emphasized specificity during algorithm development. Oxytocin was chosen as a highly sensitive yet specific criterion (ie, most postpartum women receive oxytocin and with near certainty, use of oxytocin is restricted to women during labor or immediately postpartum). Because we did not want to vaccinate women during labor, we chose a medication other than oxytocin for display of the preselected Tdap order (ie, iron supplementation). Iron is unlikely to be prescribed after an oxytocin order and before delivery (ie, during labor). Because Tdap vaccination is contraindicated for patients who have an unstable neurologic condition, an order for phenytoin during their current hospitalization blocked activation of the order; this rule excluded a single woman. A limitation of our electronic system, and for many paper-based systems, is that women might have received Tdap or Td vaccination outside of our system; therefore, we advised nurses and physicians to screen women for prior Tdap or Td vaccination or adverse reactions to Tdap before vaccine administration. Adverse reactions after Tdap vaccination are similar to those reported for the Td vaccine, most commonly pain at the injection site; serious adverse reactions occurred in less than 2% of vaccine recipients.7 During our study, we received no reports of adverse reactions resulting from Tdap vaccination.

The decision to use a pharmacy trigger (ie, iron supplementation) for display of the Tdap order was influenced by the challenges we identified during implementation of our influenza vaccination algorithm for which we used the patient discharge order.18 We believe that many failures to administer influenza vaccine were the result of the relatively short time between the discharge order and the patient's departure. For the Tdap vaccination algorithm, we were particularly concerned about the competing priorities navigated by nurses when discharging a postpartum woman; therefore, we selected a trigger that occurred earlier in the hospitalization. Although we included the discharge order as a backup trigger for women who had not been prescribed iron, only one additional woman was vaccinated. From other reports and our prior experience, we opted to present a preselected order to physicians, which minimizes their effort in accepting the order.20

The only patient factor associated with Tdap vaccine receipt was that older women were less likely to be vaccinated. This finding may have been the result of chance and needs validation in other studies. Also, because we did not interview patients, specific reasons why older women refused vaccination are unknown and elucidating these reasons would require a focused investigation.

Limitations of our findings include that generalizability is constrained by the need for an institution to have a computer decision support system; however, we expect that recent financial incentives to increase and improve electronic systems in healthcare settings will result in substantial changes.21 After installation, computer decision support systems hold promise to significantly improve other preventive interventions for hospitalized patients.22 In other hospitals such as smaller community hospitals or hospitals with sufficient personnel resources, standing orders policies and educational interventions might be as effective as our decision support strategy without the requirement for computer-based decision support systems.19,23

We observed early, dramatic, and sustained success with our computer-based decision support system strategy to increase maternal postpartum Tdap vaccination. Automated computer-based intervention strategies can dramatically improve processes of patient care such as Tdap vaccination with minimal intramural or extramural financial support. Hospitals that have comparable information system functionality should have similar success through exact reproduction or local adaptation of our algorithm.

Back to Top | Article Outline


1. Murphy TV, Slade BA, Broder KR, Kretsinger K, Tiwari T, Joyce PM. Prevention of pertussis, tetanus, and diphtheria among pregnant and postpartum women and their infants recommendations of the Advisory Committee on Immunization Practices (ACIP) [published erratum appears in MMWR Morb Mortal Wkly Rep 2008;57:723]. MMWR Recomm Rep 2008;57:1–47.
2. Lee GM, Lett S, Schauer S, LeBaron C, Murphy TV, Rusinak D, et al. Societal costs and morbidity of pertussis in adolescents and adults. Clin Infect Dis 2004;39:1572–80.
3. Cortese MM, Baughman AL, Zhang R, Srivastava PU, Wallace GS. Pertussis hospitalizations among infants in the United States, 1993 to 2004. Pediatrics 2008;121:484–92.
4. Tanaka M, Vitek CR, Pascual FB, Bisgard KM, Tate JE, Murphy TV. Trends in pertussis among infants in the United States, 1980–1999. JAMA 2003;290:2968–75.
5. Kowalzik F, Barbosa AP, Fernandes VR, Carvalho PR, Avila-Aguero ML, Goh DY, et al. Prospective multinational study of pertussis infection in hospitalized infants and their household contacts. Pediatr Infect Dis J 2007;26:238–42.
6. Bisgard KM, Pascual FB, Ehresmann KR, Miller CA, Cianfrini C, Jennings CE, et al. Infant pertussis: who was the source? Pediatr Infect Dis J 2004;23:985–9.
7. Kretsinger K, Broder KR, Cortese MM, Joyce MP, Ortega-Sanchez I, Lee GM, et al. Preventing tetanus, diphtheria, and pertussis among adults: use of tetanus toxoid, reduced diphtheria toxoid and acellular pertussis vaccine recommendations of the Advisory Committee on Immunization Practices (ACIP) and recommendation of ACIP, supported by the Healthcare Infection Control Practices Advisory Committee (HICPAC), for use of Tdap among health-care personnel. MMWR Recomm Rep 2006;55:1–37.
8. Update on immunization and pregnancy: tetanus, diphtheria, and pertussis vaccination. ACOG Committee Opinion No. 438. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114:298–400.
9. Gerbie MV, Tan TQ. Pertussis disease in new mothers: effect on young infants and strategies for prevention. Obstet Gynecol 2009;113:399–401.
10. Clark SJ, Adolphe S, Davis MM, Cowan AE, Kretsinger K. Attitudes of US obstetricians toward a combined tetanus– diphtheria–acellular pertussis vaccine for adults. Infect Dis Obstet Gynecol 2006;2006:87040.
11. Callahan ST, Cooper WO. Uninsurance and health care access among young adults in the United States. Pediatrics 2005;116:88–95.
12. Bjorkholm B, Granstrom M, Taranger J, Wahl M, Hagberg L. Influence of high titers of maternal antibody on the serologic response of infants to diphtheria vaccination at three, five and twelve months of age. Pediatr Infect Dis J 1995;14:846–50.
13. Siegrist CA. Mechanisms by which maternal antibodies influence infant vaccine responses: review of hypotheses and definition of main determinants. Vaccine 2003;21:3406–12.
14. Gerard M, Trick WE, Das K, Charles-Damte M, Murphy G, Benson IM. Use of clinical decision support to increase influenza vaccination: multi-year evolution of the system. J Am Med Inform Assoc 2008;15:776–9.
15. Hamilton LC. Statistics with Stata (updated for version 10). Belmont (CA): Cengage; 2009.
16. Cuzick J. A Wilcoxon-type test for trend. Stat Med 1985;4:87–90.
17. Winston CA, Lindley MC, Wortley PM. Lessons learned from inpatient vaccination in Michigan. Am J Med Qual 2006;21:125–33.
18. Trick WE, Das K, Gerard MN, Charles-Damte M, Murphy G, Benson I, et al. Clinical trial of standing-orders strategies to increase the inpatient influenza vaccination rate. Infect Control Hosp Epidemiol 2009;30:86–8.
19. Healy CM, Rench MA, Castagnini LA, Baker CJ. Pertussis immunization in a high-risk postpartum population. Vaccine 2009;27:5599–602.
20. Dexter PR, Perkins SM, Maharry KS, Jones K, McDonald CJ. Inpatient computer-based standing orders vs physician reminders to increase influenza and pneumococcal vaccination rates: a randomized trial. JAMA 2004;292:2366–71.
21. Blumenthal D. Stimulating the adoption of health information technology. N Engl J Med 2009;360:1477–9.
22. Overhage JM, Tierney WM, McDonald CJ. Computer reminders to implement preventive care guidelines for hospitalized patients. Arch Intern Med 1996;156:1551–6.
23. Crouse BJ, Nichol K, Peterson DC, Grimm MB. Hospital-based strategies for improving influenza vaccination rates. J Fam Pract 1994;38:258–61.
© 2010 by The American College of Obstetricians and Gynecologists.