Collaboration With School Nurses: Improving the Effectiveness of Tuberculosis Screening

Objective  To compare tuberculosis skin test (TST) reading rates between children whose tests were read by school nurses following specific requests by physicians and those who relied on their parents to get their tests read, either at school or at the physician's office.

Design  A randomized controlled trial.

Setting  An urban hospital-based pediatric practice.

Participants  Healthy low-income Hispanic and African American children aged 5 to 17 years whose physicians ordered TSTs at their routine physical examinations. Subjects attended 1 of 68 public schools. Nurses at these schools were willing to read student TSTs, and received instructions about how to read and report the results back to the physician's office.

Intervention  Subjects were randomized to a control group (routine TST placement, with no physician–to–school nurse communication) or to an intervention group (routine TST placement, with physician–to–school nurse communication).

Main Outcome Measures  Tuberculosis skin test reading rates between the 2 groups were compared. Impediments to TST reading and reporting were investigated.

Results  One hundred thirty-four children were enrolled, 54 (40%) in the control group and 80 (60%) in the intervention group. More patients in the intervention group had their TSTs read by 72 hours compared with those in the control group (74 [92%] vs 30 [56%]; P<.001). The low reading rate in the control group was best attributed to communication failures.

Conclusion  Systematic collaboration with school nurses can increase TST reading rates.

TUBERCULOSIS (TB) is a serious health problem that disproportionately affects the poor. Eighteen thousand new cases of TB occur annually in the United States, and approximately 15 million people have latent TB infection.¹ According to the Philadelphia Department of Health, the new case rate of TB in Philadelphia in 1996 was 20 cases per 100 000, with 10% of the cases occurring in children.

Our urban residency-based pediatric practice in Philadelphia is located in a high-prevalence area for TB. Many of our patients have additional TB risk factors, such as recent immigration from Latin America, close contact with prison inmates and drug-abusing adults, and homelessness. At the inception of this study, patients were screened for TB every 2 to 3 years. From September 1, 1999, to December 31, 1999, however, only 43% of school-aged patients who had tuberculosis skin tests (TSTs) placed by us had the tests read by a health professional (physician or school nurse) within 72 hours. This problem has been reported by other urban centers.^2,3

The best way to screen for TB is with the TST or the Mantoux test. The Committee on Infectious Diseases of the American Academy of Pediatrics recommends targeted screening of high-risk groups.^4,5 Before instituting targeted screening, we sought to find a way to improve the reading rate of the test in our population.

Cheng et al² tested 5 strategies to increase TST reading rates in an urban population. All groups received verbal and written instruction about the TST. The group that only received this intervention demonstrated a reading rate increase of 13 percentage points. Two other strategies, completing school forms contingent on return for test reading and dispatching nurses to children's homes to read TSTs, further increased reading rates to a statistically significant degree; however, the school form completion strategy could only be used on a few patients receiving TSTs and the home nurse visit strategy proved too costly. The other 2 strategies did not show statistically significant improvements in reading rates. The lesson learned from this study was that a new strategy should be convenient and simple, so that it is adaptable to real life and affects many of those tested.

Collaborating with school nurses to read TSTs placed at the child's routine physical examination is one such strategy. The American Academy of Pediatrics' Committee on Infectious Diseases supports using other health care professionals to read TSTs if a physician is not readily available.⁵ This article reports the results of a randomized controlled study that compared the percentage of TSTs read within 72 hours using our protocol with a more systematic approach that facilitated TST reading by school nurses.

This study was conducted at an urban hospital-based pediatric practice serving predominately low-income African American and Hispanic children. The institutional review board at the MCP Hahnemann University School of Medicine, Philadelphia, approved the study protocol. We recruited healthy children between the ages of 5 and 17 years whose physicians ordered TSTs as part of their physical examinations. Only the first child encountered from each family was included, and the child had to be accompanied by a parent or guardian. Any child receiving the TST as part of a diagnostic evaluation for an illness was excluded. Children had to be enrolled at 1 of 68 participating public schools, which included 50 elementary, 12 middle, and 6 high schools. Many schools were involved because there are more than 150 public schools within an 8.0-km radius of the clinic. Parochial and private schools were not included. All schools had a nurse on site on Wednesday, Thursday, or Friday to read the TSTs placed Monday, Tuesday, or Wednesday.

Patients were enrolled from January 10, 2000, to May 31, 2000. School nurses from all 68 schools were recruited to participate at 1 of 4 group meetings arranged to explain the study protocol, collect school information, and review the proper TST reading technique. Instruction about the TST reading technique was based on recommendations by the American Thoracic Society,⁶ as described in an article by Seibert and Bass⁷ and depicted in a Centers for Disease Control and Prevention videotape.⁸ Nurses were shown this videotape during the meeting. Of the 68 nurses, 15 did not attend the meeting. They were enrolled individually by telephone. Information and diagrams based on the article by Seibert and Bass were faxed to these nurses.

At the inception of this study, little information existed about where our patients attended school. Initially, the 40 closest schools were enrolled. After several weeks, we realized many students were not being recruited because of school affiliation. Subsequently, 28 more schools with similar demographics and close proximity to the clinic were enrolled.

The investigators (C.W.D., N.D.S., B.M., M.M.M., H.K., and L.S.) obtained written consent and interviewed parents or guardians about demographics from scripted questionnaires in either English or Spanish, with the aid of an interpreter. Eligible patients were randomly assigned to the control or intervention group by an allocation system using sealed opaque envelopes containing the group assignment. We anticipated an effect size of only 15 to 20 percentage points, necessitating a sample size of about 175 patients per group. Therefore, 350 envelopes were prepared, shuffled, numbered, and drawn in order, as each new patient presented.

After group assignment was made, instruction was given as follows: All parents were educated about the purpose of the TST and the importance of getting it read in 48 to 72 hours. Written reminders listing the TST reading date and the clinic telephone number were given to each parent. Next, each control group parent or guardian was instructed to bring his or her child back to the clinic between 8:30 AM and 4:30 PM on weekdays to have the test read or to ask the child's school nurse to read the test and call our office with the result. Instructions given to control group parents were the same as those given to all parents or guardians of school-aged children who received TSTs at our clinic before and during the study period. We provided some additional information to this group while explaining the purpose of the study and obtaining consent.

Intervention group parents or guardians received the same information as control group parents with the exception of test-reading instructions: TSTs were to be read only by the school nurse. Parents or guardians gave us permission to notify the nurse about test placement by facsimile or telephone and to request test reading in 48 to 72 hours. Nurses were instructed to report the results directly to us on the day the test was read. Any child noted to have an indurated TST result was told to return to our office after school for an official reading by a physician. Nurses were asked to report the degree of induration in millimeters. Negative test results were reported as 0 mm.

Parents or guardians of children who did not have their tests read were called within 1 week of test placement. A scripted interview was used for this call. School nurses were also called if the parent could not explain why the test was not read.

We assumed a 43% return rate (based on the return reading rate 3 months before the study commenced) and a Hawthorne effect between 10 and 15 percentage points when calculating the sample size. The Hawthorne effect estimate was based on the increase in the TST reading rate observed by other investigators² when education was introduced as a strategy for a similar population. We estimated our control group reading rate would increase to about 55% once the study commenced because the enrollment process might influence behavior. Initially, we determined we would need to enroll 162 patients per group to give us 80% power to detect an effect size of 15 percentage points. Four months into the study, we recognized that the effect size was closer to 30 percentage points. We recalculated the number of subjects needed to give us 80% power to detect an effect size of 30 percentage points (increasing the return reading rate from 55% to 85%) for a 2-tailed test, α = .05, using the z statistic to compare proportions of dichotomous values; this number was about 35 per group.

Demographic characteristics were compared by calculating the means and medians of the children's ages or by calculating the frequency distribution and χ² of the characteristics listed in Table 1 using computer software (Epi-Info 6; Centers for Disease Control and Prevention, Atlanta, Ga). Comparison of the TST reading rates between the control and the intervention groups was analyzed using the χ² test. Telephone call responses were examined to identify common reasons for failing to have tests read.

When we first began recruiting patients, we were bound to draw envelopes in numerical order as dictated by the randomization process described. After several months, it became apparent that the effect size was actually 30 percentage points. Statistically, only 35 patients are required in each group to demonstrate an effect of 30 percentage points above the anticipated baseline reading rate of 55%. Therefore, we made the decision to complete the study at the end of the school year, because we attained an adequate sample size in each group. At that point, the distribution of randomly assigned patients was unequal. If we continued to recruit all 350 patients, we would have finished with equal numbers in each group.

Six hundred thirty-four school-aged children presented to the clinic and had TSTs placed during the study period. Of these children, 157 (25%) were eligible for the study and 143 (23%) were recruited. Four parents refused to participate. Five patients were dropped from the study: 2 were enrolled erroneously, and 3 control group school nurses were accidentally sent facsimiles. Final enrollment was 134 children (21%), with 54 (40%) in the control group and 80 (60%) in the intervention group. Patients were excluded from the study for various reasons (Figure 1). Many patients were ineligible because of school affiliation. Although we tried to involve enough schools, we quickly learned that our patients hailed from a wide geographic area involving many schools. In retrospect, we needed to enroll twice as many schools to capture most patients.

Fourteen eligible patients were missed. Three parents left before consent could be obtained. Eleven patients were not recognized as attending eligible schools. In 3 cases, the parent could not pronounce or spell the school name. Eight others were missed because of staff oversight.

The children's ages ranged from 5 to 17 years. Participants in each group did not differ for age, sex, race, insurance status, household member with a previously positive TST result, mode of transportation to the clinic, single-parent status, or employment status of the primary caretaker (Table 1).

The TST reading rate for school-aged children during the 4 months preceding patient enrollment was 43%. During the study period, the reading rate increased to 56% in the control group, and to 93% in the intervention group. This difference was statistically significant (P<.001) (Figure 2).

Of the 28 control group parents whose children did not have documented TST results, 27 were contacted. Four tests were later verified as being read by the school nurse. Of the remaining 23, all but one of those contacted intended to have the school nurse read their child's test. Seven tests were not read because of school absence. In one other case, the child did not go to the nurse when summoned. Thirteen others were not read because the parents failed to effectively communicate their wishes to the school nurses. Two other parents made no effort to get the tests read.

All of the school nurses of the 11 intervention group patients who did not have a documented TST result after 96 hours were called. Four of these patients actually had the test read by 72 hours, but the result was not communicated back to the investigators because of technical problems. One other child in the intervention group had the test read at the right time but had it read at our clinic instead of school. Of the remaining 6 patients, 3 tests were not read because of school absence, 2 tests were not read because the student did not go to the nurse's office when summoned, and 1 nurse received the facsimile 72 hours after test placement.

During the study period, 8 of the 634 children had positive TST results (1 in the control group, 1 in the intervention group, and 6 in the group deemed ineligible for the study) using 10 mm or greater as the cut point.⁹ Three of these children had risk factors (all in the nonstudy group); one had a household contact with an active case of TB. Because the reading rate in the nonstudy group was only 40%, it is difficult to estimate the true incidence of TST conversion.

Tuberculosis screening of high-risk populations is considered one of the essential components of a successful TB control program.¹⁰ Unfortunately, TB screening efforts in urban areas are plagued with low turnouts for TST reading. Health care professionals working in high-risk areas are encouraged to collaborate with local TB control programs and each other to enhance TB screening efforts.^5,11

During the enrollment process, almost all of the control group parents wanted the school nurse to read the TSTs. Their behavior supported this preference, because only 3 of 54 control group patients returned to the clinic for test reading. Parent–to–school nurse communication failures were primarily responsible for lack of TST reading or reporting. Our efforts to streamline communication between physicians and school nurses greatly improved TST reading rates in the intervention group. Other benefits of this intervention included the following: (1) increased education of school nurses about TB; (2) improved school nurse awareness of proper test reading and reporting; (3) more accountability for reporting and recording TST results, especially in a large-volume clinic; and (4) better planning to read TSTs, because we identified days when school nurses could read the tests.

We did not test the school nurses' ability to read TSTs accurately because we did not ask them to decide which tests were positive. Patients with any degree of induration were referred back to us within 24 hours for TST reading, examination, and treatment, if necessary. All 5 study patients referred back to the clinic were seen within 24 hours. Nurses had a tendency to overread induration. This contradicts the findings of Kendig et al,¹² possibly because the circumstances for TST reading by school nurses in this intervention were different from those imposed by their study.

Several biases were present in the study design. Two of these potentially increased the intervention group reading rate. First, the study was not blinded. School nurses receiving facsimiles to read TSTs knew these were intervention patients and may have been more conscientious about reading and reporting results. This does not seem to be the case, however, because control group parents relied heavily on school nurses to read their children's tests. When this was effectively communicated to the nurse and the child was not absent, the tests were read in all cases but 1 in the control group and 2 in the intervention group.

A second bias was introduced by contacting intervention group nurses after sending them their first facsimile. While this helped ensure the functionality of facsimile machines, it was potentially an extra reminder. Eighty patients were enrolled in the intervention group from 35 different schools; therefore, 45 intervention group nurses did not receive a telephone call in addition to the facsimile. Of this group, 40 patients (89%) had their TSTs read on time. Thus, the additional telephone call did not influence the intervention group results enough to alter our conclusions.

Last, was bias introduced by limiting patient enrollment to specific schools? Parochial, private, charter, and home schools were excluded. Parents sending their children to these schools might respond differently to the charge of getting TSTs read on time. The reading rate of the 43 patients attending these schools was 42%; exclusion of this group did not seem to introduce bias.

Limiting recruitment to children attending schools closest to our clinic may have introduced bias if transportation or convenience influenced the test-reading rate. Selective student enrollment from schools closest to the clinic might skew the control group reading rate to be higher than average. The converse would be a lower than average reading rate among ineligible patients. Indeed, the return rate for patients who were ineligible for the study because they attended nonparticipating public schools during the study period was 40%, slightly lower than the baseline of 43% observed before the study commenced.

One of the strengths of this program is its real-life applicability. Many urban outpatient clinics already rely on school nurses to help read TSTs (Harriet Weinstein, RN, oral communication, March 9, 2000). Nurse recruitment is not difficult if group meetings can be arranged. A designated telephone line for nurses to report results also improves the communication process. Despite the lack of year-round school nurse availability, this intervention is worthwhile for inner-city youth of low socioeconomic status, and can be expanded to preschool-aged children attending programs in which nurses or other health care professionals can be instructed to identify children with indurated TST results. This intervention is adaptable to e-mail, if patient confidentiality can be assured.

Collaboration with health care professionals with access to children successfully "bridges the gap" to improve TST reading rates. A physician's ability to determine whom to screen is preserved; reading is more convenient for parents or guardians; and verification, management, and treatment of positive test results reverts back to the primary physician.

Accepted for publication June 21, 2001.

Presented (as a final requirement for a Primary Care Faculty Development Fellowship) at the Michigan State Primary Care Research and Development Conference, East Lansing, June 8, 2000; and as a poster at the Societies for Pediatric Research/Ambulatory Pediatric Association Spring Meeting, Baltimore, Md, April 30, 2001.

We thank the Michigan State Primary Care Faculty Development Fellowship Program's faculty for their critical appraisal and expertise.

Screening patients with risk factors for TB with the TST or the Mantoux test is recommended; however, the effectiveness of the TST is of no value if it is not read. Our experience and the experience of others working in urban practices where many patients have risk factors is that only half of the TSTs placed are read. Interventions to improve reading rates have been minimally successful.

This study provides clinicians with another way to improve TST reading rates. We found that school nurses are reliable resources for reading TSTs once their roles are clearly assigned and communication is streamlined.

Corresponding author and reprints: Cynthia W. DeLago, MD, Section of General Pediatrics, St Christopher's Hospital for Children and MCP Hahnemann University School of Medicine, Erie Avenue at Front Street, Philadelphia, PA 19134-1095 (e-mail: Cynthia.W.DeLago@drexel.edu).