Case Study 3: A Cost Benefit Analysis for Dissemination of an Evidenced-Based Program

R. Windsor, “Smoking Cessation and Reduction In Pregnancy Treatment (SCRIPT) Methods: A Meta-Evaluation of the Impact of Dissemination,” American Journal of Medical Sciences (October 2003), 216-222 (see publication for references).

Active and passive exposure to tobacco smoke during pregnancy and infancy are the most serious and preventable causes of adverse maternal, fetal, and infant outcomes in the United States. This meta-evaluation provided a synopsis of the state of the science in five areas of this specialized area: (1) the validity of patient reports of smoking status and trends during pregnancy: (2) a definition of “best practice” cessation methods for pregnant women; (3) a description of the cost of the SCRIPT methods (niput); (4) an estimate of the impact of dissemination of the Agency for HealthCare Research and Quality (AHRQ, 2000) recommended SCRIPT methods among the 800,000 pregnant US smokers in 2002 (output); and (5) an estimate of the evidence from a cost-benefit analysis of improved maternal and infant outcomes from cessation attributable to SCRIPT methods, including a Sensitivity Analysis (ouput).

Defining the Problem and Population at Risk

Approximately 4 to 4.2 million women gave birth in the United States from 1990 to 2003: about 1.6 million (40%) were Medicaid patients. During this period, the smoking prevalence rate during pregnancy had been monitored annually by federal agencies. Data in Table 6.7 are based on the PRAMS (Pregnancy Related Assessment and Monitoring System) reports from the CDC National Center for Health Statistics.

All smoking rates are based on self-reports in response to a multistage, mailed survey (70% response rate) sent two to four months after babies were born. The CDC reported that the rate during pregnancy had been reduced from 18.4% in 1990 (736,000 smokers) to 12.2% in 2000 (488,000 smokers) to 10% in 2003 (410,000 smokers). Thus, the CDC

table 6.7 Smoking Rates During Pregnancy and Percent Change by Group: 1990-2003

years

total

black

white

H ISPANIC

oth er

1990

18.4%

1993

15.8%

15.9%

21.0%

6.7%

8.6%

1995

13.9%

1997

13.2%

10.6%

17.1%

4.3%

7.5%

2000

12.2%

9.1%

15.7%

3.7%

5.9%

2003

10.0%

% < 45% < 320,000 smokers = self-reported estimate Use 1 < only reports indicated that approximately 320,000 more smokers quit during their pregnancy in 2003, compared to the 1990 data: a > 40% reduction in the annual smoking rate during pregnancy.

Data in Table 6.8 documented the average smoking rate for three-year periods among pregnant women from 1990 to 2003 based on the National Annual Household Survey on Drug Abuse by the Substance Abuse and Mental Health Services Administration (SAMHSA). These data represent a subsample of women, ages 15-44, from the annual 70,000 face-to-face interviews in the home, also with a 70% response rate. The SAMHSA data do not agree with the CDC data. The SAMSHA interviews, however, were much more likely to have elicited a valid statement of smoking status compared to a survey mailed two to four months after birth.

Because NONE of the CDC or SAMSHA surveys used a biochemical test (e.g., a saliva or urine cotinine analysis) to corroborate patient self-reports, all national estimates of the prevalence and number are significant underestimates of the problem. Because of the high social desirability of a nonsmoking response, high rates of patient deception/ nondisclosure about smoking status at the onset and during care have been well documented. The CDC documented a deception rate of 48% by a urine cotinine test (N = 6,000), and the Alabama Smoking Cessation/ Reduction In Pregnancy Treatment Trial II documented a deception rate 24% by a saliva cotinine test (N = 1000). If the SAMHSA rate of 17.0% is adjusted, by adding a low estimate of deception (5%), the true prevalence rate was > 20% (> 800,000 smokers) in 2003. When the CDC and SAMHSA data are considered, it is a plausible conclusion that the CDC prevalence rates are very inaccurate.

Assessing Smoking Status and Exposure in Prenatal Care

In 1991 in the Journal of the American Medical Association, Fiore presented a discussion on the need to routinely assess smoking status

table 6.8 Average Smoking Rates, Women 15-44: 1990-2003

years

pregnant

not pregnant

1990-1992

20.0%

30.0%

1994-1996

20.6%

31.8%

1998-2000

19.4%

30.2%

2001-2003

17.5%

30.0%

and recommended its inclusion as a standard procedure in clinical practice: “a new vital sign.” The Agency for Healthcare Research and Quality Clinical Practice Guideline (2000) recommended routine assessment of patients: “The first step in intervention is assessment of tobacco use status. Biochemical confirmation was recommended.”

Defining the Effectiveness of Health Education Methods for

Pregnant Smokers

The PHASE 1 and 2 SCRIPT evaluations, cited by the Clinical Practice Guidelines, documented the quality, efficacy, and cost effectiveness of standardized patient education for pregnant smokers: SCRIPT internal validity. According to the Guidelines, “clinicians should offer effective smoking cessation interventions to pregnant smokers at the first prenatal visit as well as throughout the course of pregnancy.” Examples of effective interventions with pregnant smokers from the meta-analysis and Guidelines (p. 94) are presented in the article.

Defining the SCRIPT Program for Dissemination

Phase I, II, and III Trials (1982-2002) and meta-evaluation of the eight studies among 2,700 patients from four countries (US, Canada, Sweden, and Norway) comprehensively evaluated the SCRIPT Program. A synopsis of the SCRIPT program is presented below

SCRIPT ASSIST Component

The ASSIST component of SCRIPT, the core methods, includes a three-part patient education process at the first visit.

Component #1: “Commit to Quit Smoking—During and After Pregnancy,” a 10-minute video, was designed to enhance motivation to quit, improve comprehension of risk information, introduce the Guide, ensure exposure to recommended cessation methods, and reduce counseling time.

Component #2: “A Pregnant Woman’s Guide to Quit Smoking” is a 32-page tailored guide with a fifth- to sixth-grade reading level. It is introduced by the video and the clinic counselor.

Component #3: This is a 5-7-minute patient-centered counseling session delivered by a physician, nurse, midwife, social worker, or WIC nutritionist during regular prenatal care.

The primary cost of SCRIPT methods is staff time and patient education materials. Because an agency perspective was applied, patient time, facilities, and intervention development costs were not used in the cost analysis. An RN would be a typical SCRIPT provider. An average salary + fringe of $60,000-80,000/year can be used to estimate nursing staff costs to estimate intervention costs in 2015 for in-class discussion and computation of costs/CEA/CBA.

An average staff cost for delivery of the 10-minute intervention was $5.00/patient. The bulk order (1,000 +) price for “A Pregnant Woman’s Guide to Quit Smoking” is approximately $3.00. The commercial price of the “Commit to Quit Smoking” video is $25.00/unit. The video cost can be divided by 100 patients/year for a cost of $0.25/patient for clinical use, not for individual or commercial distribution. Because a television and DVD player are standard equipment at a clinic, they were not included in the cost estimates. Total intervention cost was $8.00/patient.

Patient flow analyses at the prenatal care sites would confirm that nurses, with reinforcement by other staff, would spend about 3-5 minutes with smokers. C group patients typically received “Ask and Advise” procedures. Nurses provided printed handouts on the risks of smoking and the benefits of quitting. The average total new time needed would be about 6-8 minutes. The costs associated with staff time to deliver the usual cessation advice are about $1-2 per C group patient. Thus, the approximate NEW cost to an agency of the SCRIPT methods would be about $6.00/ patient.

Estimated Behavioral Impact of SCRIPT Dissemination

Table 6.9 presents a synopsis of the impact of the eight SCRIPT evaluation studies.

The evaluations in Table 6.9 documented, by a self-report and biochemical measure, the effectiveness of SCRIPT methods. As noted, the overall average behavioral impact level was 7.7%, and US behavioral impact level was 8.5%. Table 6.10 presents estimates of the impact that can be used to conduct a CBA of national dissemination of SCRIPT methods. It applies three dissemination impact levels: low = 4%; moderate = 5%; high = 6%, to estimate the additional number and percentage of patients who might quit from SCRIPT methods. As indicated in Table 6.7, exposure among the 700,000 smokers (17% X 4 million +) to effective methods might produce annually an additional 4% cessation rate (low impact N = 28,000

pi: evaluation study

м easure

e group

c group

difference

n

%

n

%

(E-c)

Windsor, 2000 (US)

S-COT

139

17.3%

126

8.8%

8.5%

Gebauer, 1998 (US)

S-COT

84

15.5%

94

0.0%

15.5%

Hartmann, 1996 (US)

CO

107

20.0%

100

10.0%

10.0%

Valbo, 1996 (Norway)

CO

107

27.0%

105

11.4%

15.7%

Windsor, 1993 (US)

S-COT

400

14.2%

414

8.4%

5.8%

O’Connor, 1992 (Canada)

U-COT

90

13.3%

84

6.0%

7.3%

HJfrson, 1991 (Sweden)

SCN

444

12.6%

209

8.6%

4.0%

Windsor, 1985 (US)

SCN

102

13.7%

104

1.9%

11.8%

US Studies (N = 1,670)

Total

15.4%

6.9%

8.5%

Non-US Studies

Total

15.0%

8.8%

6.2%

(N = 1,039)

(N = 2,709)

Total

15.2%

7.5%

7.7%

E minus C Group difference = 4.0% to 15.7%

table 6.10 Estimated Smokers and Impact of AHRQ-SCRIPT Methods

level-cpd

smokers

(a)

standard m ethods rate

impact

ahrq-scri pt m ethods rate (b)

impact

difference

(b-a)

Light (10)

400,000

8%

32,000

16%

64,000

32,000

Mod

200,000

6%

12,000

12%

24,000

12,000

(11-19) Heavy (20)

100,000

2%

2,000

4%

4,000

2,000

Total

700,000

6.2%

46,000

12.5%

92,000

+ 46,000

quitters: cohort 1), an additional 5.0% cessation rate (moderate impact N = 35,000 quitters: cohort 2), or an additional 6% cessation rate (high impact N = 42,000 quitters: cohort 3).

The estimated cost to deliver the SCRIPT methods to all pregnant smokers would be $8.00/patient in 2003. Thus, the cost for a pregnant smoker cohort would be about $5.6 million ($8.00 X 700K). The cost of SCRIPT can be varied to $10.00/patient ($7.0M) and $12.00/patient ($8.4M) to reflect regional cost variations. With these assumptions, a CBA and sensitivity analysis, describing the cost-benefit of nationwide SCRIPT dissemination, can be performed.

Using the excess smoking-attributable savings of $1,500/patient who quit in 1995 (Miller et al., 2001), and adjusting this cost for inflation by the Consumer Price Index (CPI)—US Bureau of Labor Statistics (BLS) 314 | Evaluation of HP-DP Programs

cohort

impact

quit

rate

savings

scri pt cost

n et savings range

#1 700K

4%

28,000

$56M

$5.6M

$50M-$49M-$48M

#2 700K

5%

35,000

$70M

$7.0M

$64M-$63M-$62M

#3 700K

6%

42,000

$84M

$8.4M

$78M-$77M-$76M

M = millions

(A)

(B)

(C)

(B1-2-3... C1-2-

$2,000

3... D1-2-3)

for increases in maternal and infant healthcare costs, in January 1, 2003, the average estimated savings for each quitter would > $2,000. When the savings from the excess smoking-attributable care costs of $2,000/ patient is multiplied by each impact level (column A = $2,000 X column B), the potential benefits of dissemination is $56 million (low effect cohort 1), $70 million (moderate effect cohort 2), and $84 million (high effect cohort 3). CBA data in Table 6.11 indicate that dissemination is cost-beneficial.

When sensitivity analyses are performed, varying and subtracting low-moderate-high estimates of SCRIPT costs ($8.00 = $5.6 million, $10.00 = $7 million, and $12.00 = $8.4 million per patient (column C) from the low-moderate-high average estimates of gross savings (column B), the estimates net savings range (column D) would be $49 million, $63 million, and $77 million per year. If the lowest estimate of behavioral impact (4%/28,000) and highest estimate of intervention costs ($12/$8.4M) are applied in a CBA, the CBR = $1: $5.7 ($47.6 M/$8.4 M). If the impact from SCRIPT dissemination is set at 5% (+ 35,000 quitters), the estimate of excess smoking attributable healthcare cost is reduced 25%, $2,000 to $1,500 (35,000 X $1,500 = $52.5M), and the highest SCRIPT cost of $12/ patient (700K X $12 = $8.4M) is applied in a sensitivity analysis, the estimated savings from dissemination would be $27 million ($35M/$8.4 M), and ROI = $1: $4.17.

 
Source
< Prev   CONTENTS   Source   Next >