Health Education Research, Vol. 15, No. 4, 469-483,
August 2000
© 2000 Oxford University Press
Process measures in an antenatal smoking cessation trial: another part of the picture
Discipline of Behavioural Science in Relation to Medicine, Faculty of Medicine and Health Sciences, University of Newcastle, NSW 2308, and
1 Department of Obstetricsand Gynaecology, John Hunter Hospital,New Lambton Heights, NSW 2305, Australia
Correspondence to: R. Walsh, New South Wales Cancer Education Research Program, Locked Bag No. 10, Wallsend, NSW 2287, Australia
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionConclusionReferences |
|---|
Data on provider and patient compliance can be crucial in understanding the degree of a health education program's effectiveness, as well as in identifying areas where the program requires modification. However, such data are rarely systematically reported in randomized trials. This report assesses the degree to which doctors and midwives complied with intervention protocols in a hospital antenatal smoking cessation trial, and also examines the program's acceptability to patients. Provider compliance was assessed principally via consultation audiotapes and provider-completed checklists. The audiotape analysis identified substantial compliance problems. For example, in relation to six specific smoking-related pregnancy risks, the proportions of Experimental Women informed about each individual risk ranged from 26 to 38% and the proportions receiving counselling items ranged from 52 to 79%. Doctors only informed a minority of Experimental Women of the increased risk of Sudden Infant Death Syndrome (28%) and of the presence of toxic chemicals in tobacco (21%). Comparison of compliance data from audiotapes and provider checklists revealed there was no significant agreement in three of four cases tested. Experimental Patients completed questionnaires to assess recall of smoking advice and to rate 12 program features. Of specific Experimental Program elements, the videotape (85%) received the highest level of positive patient ratings and the lottery (42%) the lowest. The process evaluation indicated that the Experimental Program needed some modification to increase its suitability for routine application. The findings also support the value of including an objective measure of provider compliance.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
The efficacy of an intervention when applied in the context of a controlled trial may be substantially reduced when introduced into routine health services (Lichtenstein, 1997
). Health providers fail to provide routinely many preventive interventions of proven efficacy (Kottke et al., 1990). In smoking cessation research, better quality information can be collected and more useful inferences made when the intervention is implemented by professionals who would normally administer it (Owen, 1989). However, difficulties can arise when a research team interacts with primary care staff (Doherty et al., 1988). Competing pressures and unrealistic expectations can lead to loss of provider enthusiasm and commitment. This, in turn, can result in non-compliance problems such as low recruitment rates and poor intervention coverage (Doherty et al., 1988; Wadland et al., 1990). Despite the significance of these issues, data on provider and patient compliance and acceptability are reported infrequently (Walsh and Redman, 1993). This is probably due to a number of factors such as preoccupation with study outcomes, sensitivities of research and clinic staff or failure to recognize the importance of compliance issues in determining a program's effectiveness.
Type of compliance problems
When compliance levels have been measured in smoking cessation studies conducted outside of the antenatal setting, three major types of problems have been identified: a tendency for control groups to receive higher levels of care than specified, e.g. Thompson et al. (Thompson et al., 1988) noted there was evidence that some physicians `did the same thing on every patient'; a general low rate of adherence to the experimental group protocol, e.g. Li et al. (Li et al., 1984) noted that 34% of patients in one clinic did not receive the physician stop smoking message and Russell et al. (Russell et al., 1987) indicated that compliance averaged 45% and fell during the study year; and a very low rate of compliance with specific elements of experimental interventions, e.g. making an appointment to follow-up on smoking advice appears to be a particularly problematic area for doctors with rates of 15% (Cummings et al., 1989a), 19% (Cummings et al., 1989b) and 10% (Kottke et al., 1989) in three trials. Asking patients to set a quit date is another intervention element which doctors appear to have difficulty with. Patient recall of quit date discussion in the three studies above was 38% (Cummings et al., 1989a), 38% (Cummings et al., 1989b) and 18% (Kottke et al., 1989).
A number of possible methods exist to measure compliance and other aspects of process in smoking cessation research. The main measures are provider-completed checklists, patient recall of the intervention and direct observation techniques such as videotape or audiotape (Walsh and Redman, 1993). All of these measures have some methodological limitations.
For example, in relation to provider-generated data, evidence supports the view that doctors consistently overstate their level of preventive care provision when compared against other measures such as patient exit interviews or medical record data (Lewis, 1988). Patient recall has also been found to be systematically biased towards over-reporting by smokers of advice to quit (Ward and Sanson-Fisher, 1996). Although Folsom and Grimm (Folsom and Grimm, 1987) found patients under-reported smoking advice, their study involved a smaller sample size, a less powerful methodology and more distant patient recall. In the case of direct observation techniques, more objective data are produced. However, there is debate about the intrusiveness of direct observation and about the degree of reactivity which may occur in observed consultations (Redman et al., 1989).
Criteria for suitable programs
It has been recommended that five criteria must be met for a smoking cessation program targeting pregnant women to be suitable for routine application by usual care providers (Walsh and Redman, 1993). These criteria were developed by drawing on the seminal research in the area of patient education and compliance by Haynes et al. (Haynes et al., 1979) and the findings of the later review of smoking cessation intervention research among pregnant women by Windsor and Orleans (Windsor and Orleans, 1986). In their review, Walsh and Redman (Walsh and Redman, 1993) also analysed the results of 20 controlled evaluations of smoking cessation interventions in pregnancy before proposing the five recommended criteria as follows: (1) program materials must be readily available, (2) time commitment must be feasible for the provider, (3) training required to use the program must be clearly described and of acceptable duration, (4) program components must be acceptable to providers, and (5) program components must be acceptable to patients. No antenatal studies have reported acceptable data on any of the last three criteria (Walsh and Redman, 1993). In assessments of any preventive program, it is vitally important to monitor whether health providers have delivered the intervention as planned. In antenatal smoking research, it has been emphasized that the problem of staff compliance in randomized trials is as important as patient compliance (Lumley, 1987). In one British study, only 10% of experimental patients recalled receiving the intervention in full and 10% did not receive either of the intervention elements: advice from the obstetrician and a leaflet (MacArthur et al., 1987). In another trial, Shakespeare (Shakespeare, 1990) reported that only just over half of all eligible women were offered the full intervention.
Need to monitor compliance
Self-report surveys of health care providers' practices and views about smoking cessation interventions in pregnancy also support the need for provider compliance to be monitored in trials in this area. For example, a survey of Australian antenatal clinic directors found that only 57% of medical directors and 79% of nursing directors believed that advice on smoking was essential at the first antenatal visit (Walsh et al., 1995). In the same survey, the directors estimated that only about one-third of smokers attending their clinics were counselled about methods of quitting and less than 10% were encouraged to set a quit date (Walsh et al., 1995). Most general practitioners, midwives and obstetricians providing antenatal care in a British study said they experienced difficulty and a lack of enjoyment while giving smoking cessation counselling, and only 28% thought that they possessed the necessary skills (Clasper and White, 1995). Clearly, these results which demonstrate a low priority attached to smoking cessation intervention by antenatal care providers and low confidence levels, suggest that provider compliance problems are likely to be major issues to confront researchers in this area.
It is not surprising that in a recent metaevaluation, Windsor et al. (Windsor et al., 1998) argued that process evaluation data were critical in reports of smoking cessation interventions in pregnancy. They highlight the disappointing finding that, in most smoking cessation trials in pregnancy, process evaluation data are nearly absent. Windsor et al. (Windsor et al., 1998) point out that data on the level of exposure to each intervention component delivered to patients is critical to make a conclusion about behavioural efficacy and internal validity. This point becomes especially crucial as the field moves from efficacy trials where interventions are delivered by health education specialists affiliated with the research team (Sexon and Hebel, 1984) to effectiveness trials where interventions are delivered by usual care staff (Hjalmarson et al., 1991). These comments also apply to many other health education fields such as cancer control programmes, where there is increasing emphasis on evaluations of interventions by usual care providers (Glynn and Cullen, 1989).
The present report focuses on an antenatal smoking cessation trial with previously published outcome data (Walsh et al., 1997). In the trial, recruitment biases were not evident since there were no significant differences between the Experimental and Control Groups on any of the 12 baseline variables tested. Table I summarizes the biochemically validated point prevalence abstinence rates reported in the trial. For these analyses, women lost to follow-up were counted as failures.
|
Although significantly different at all three measurement points, the cessation rate increase in the Experimental Group is modest at end-of-pregnancy (7%). Using a consecutive cessation criterion, i.e. validated abstinence at both the midpoint and end-of-pregnancy, the difference is 9% in favour of the Experimental Group. In a meta-analysis, Dolan-Mullen et al. (Dolan-Mullen et al., 1994
) found that antenatal smoking interventions had a combined risk ratio of cessation of 1.50 (95% CI 1.22–1.86). It is difficult to compare this trial with most studies synthesized by Dolan-Mullen et al. (Dolan-Mullen et al., 1994) because nearly all were delivered by specialist educators. However, the quit rates achieved are similar to those quoted in the meta-analysis.
This paper focuses on the process evaluation of the Walsh et al. (Walsh et al., 1997) trial rather than the outcome data and has two main objectives: (1) to measure the degree to which doctors and midwives complied with the intervention protocols, and (2) to document the extent to which the Experimental Program satisfies published criteria for assessing whether a program is suitable for routine application by health care providers.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Recruitment and intervention conditions
The study was undertaken in the period 1990–92 at a teaching hospital in Newcastle, Australia. The recruitment procedures and intervention conditions have been described elsewhere (Walsh et al., 1997
). Summarizing, smoking women attending for their first visit at a public antenatal clinic and less than 26 weeks gestation were randomized either to a control condition or a cognitive-behavioural smoking cessation intervention by clinic staff. The midwives who recruited women into the trial conducted the normal booking interview at the clinic where patients were assessed prior to seeing their allocated doctor. In all, 27 midwives were responsible for recruiting one or more women into the study. Approximately 18 doctors were involved in the study, and their backgrounds included visiting and staff obstetricians, registrars and junior medical officers. The clinic was characterized by a substantial amount of staff rotation.
The Control Protocol required both the doctor and midwife to inform patients that smoking was an important cause of problems in pregnancy and that they should stop smoking. Control Patients were also to be given a sticker, a smoking risks for women pamphlet and a short cessation guide. None of these materials were tailored specifically for pregnant women. Prior to the study, clinic staff indicated the control protocol was equivalent to their optimal level of usual care. Pre-study usual care did not include discussion of specific risks of smoking in pregnancy or of methods of quitting smoking.
The Experimental Protocol included: 14 min videotape, patient manual with a quit date form, offer of four packets of sugarless chewing gum, attempted involvement of a support person with a special letter and contract, patient follow-up letter, quitters' lottery for those not smoking at their second visit, plus doctor advice and midwife counselling by usual clinic staff.
The doctor protocol required 2–3 min of standardized risk information during the first consultation with brief follow-up at the second visit and at 34–36 weeks gestation. The first visit component included mention of six specific, pregnancy-related smoking risks. The midwife protocol required approximately 10 min, and followed the same format as the videotape and manual, i.e. risk information then countering of exemptions and finally cessation strategies. The interaction was complemented by a flipchart to be shown to patients. Risk information focused on the same six risks as that contained in the medical protocol. The counselling section included 14 specific items: nine strategies to assist patients during the first stage of quitting and five tips (the 5Ds) to help cope with any subsequent cravings. Negotiation of a quit date was also part of their protocol. Midwives were also responsible for arranging the videotape screening, distributing the manual, offering the confectionery gum, incorporating the support person procedure and informing women about the lottery. Similar to the doctor protocol, the nurse protocol also called for follow-up by the midwife tailored to the patients' quit status, at their second visit and 34–36 weeks visit. Table II summarizes components of the Control and Experimental Interventions.
|
Provider training
A 45 min training session included an outline of the protocol for the trial, a detailed description of the behavioural program and open discussion. Midwives also spent 30 min practising delivery of the counselling component of the program with a member of the research team. A research assistant who visited the clinic approximately three times each week trained new staff, and assisted medical and nursing staff experiencing difficulties with the program.
Provider compliance and acceptability
Provider compliance and acceptability was assessed using two main measures: audiotapes of consultations and provider-completed checklists. Direct observation such as audiotaping is an effective approach to monitoring provider–patient interactions (Gordon et al., 1989). Patient recall of smoking advice and ratings of program elements also provided broad information about the delivery of intervention components.
Prior to the study, a target of audiotaping 25% of doctor and midwife components was established. A convenience sampling method was used. The research assistant was instructed to obtain recordings throughout the study period from as many clinic staff as possible. Emphasis was placed on recording more Experimental Components than Control Components. Recordings were mainly made on days when the research assistant had time to prompt staff. The audio recordings were rated by a trained rater, who was blinded to the patients' group allocation. Reliability checks were undertaken during the rating process. A random selection of audiotapes were rated by a second rater to assess levels of agreement. For the risk information section of the midwife and doctor components, a score out of 6 was calculated based on the number of specific risks mentioned on each recorded patient's audiotape. A similar procedure was adopted to determine a midwife counselling score out of 14. The counselling score included the nine strategies and the five tips.
After the first consultation with Experimental Group patients, doctors and midwives were asked to complete a checklist indicating which elements of the protocol they had delivered. The main section of the midwife and doctor checklists listed 19 and eight intervention items, respectively, and for each item required the circling of a `Yes', `No' or `Not Appropriate' option. The checklists also asked the providers to rate aspects of the acceptability of using the Experimental Program with the previous patient on a five-point scale.
Provider time commitment
The audio recordings of the doctor and midwife components were timed by stopwatch to establish the duration of these intervention components.
Patient recall of smoking advice and acceptability
Rates of withdrawal of consent were assessed in both groups of patients as one measure of acceptability. The self-administered midpoint questionnaire asked patients in both groups about their recall of smoking advice. Patients who recalled receiving advice about how to stop smoking were asked to rate the helpfulness of this advice. In addition, Experimental Group patients were asked to rate 12 other general and specific features of the Experimental Program.
Analysis
Continuity-adjusted 
2 were used to test for significant differences between categorical data. Fisher's exact test was used where there were small expected frequencies. The t-tests were used for measurement data. The 
statistic and Z scores were used to test degree of agreement of categorical variables.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Recruitment and outcome data have already been reported (Walsh et al., 1997
). In summary, a 53% consent rate was achieved and final patient numbers in the two conditions were: Control Group = 125 and Experimental Group = 127. Recruitment rates varied significantly between different clinic midwives.
Provider compliance and acceptability
Audiotape data
Audio recordings were made of 25% of midwife components (33% of Experimental Group and 18% of Control Group) and 27% of doctor components (42% of Experimental Group and 22% of Control Group). No doctors or midwives refused the request to make an audiotape. In relation to the reliability checks, could not be calculated because of the relatively small number of audiotapes (n = 18) re-rated. However, for each audio recording, across all intervention content items, over 80% agreement was found.
Midwife adherence to the risk information and counselling sections of their protocol based on ratings of the audio recordings is outlined in Table III. In relation to the six risk information items, there was no significant difference on any item between the proportion of the Experimental Group who received advice from the midwives and the proportion of the Control Group. Approximately one-third of the women in both groups received information from the midwives on each risk information item. Consistent with this finding, there was also no significant difference between the midwives' mean risk information score for Control Group patients (1.54) and Experimental Group patients (1.95).
|
Midwife compliance with the counselling section of the Experimental Program was higher than with the risk information section. The proportion of Experimental Group patients receiving counselling items ranged from 52 to 71%. Contamination was documented in one recording where a Control Group patient received the Experimental intervention. On every item, the proportion of women in the Experimental Group who received counselling was significantly higher than the proportion of women in the Control Group. In addition, the mean counselling score was significantly higher (t-test, P < 0.0001) for Experimental Group patients (9.02) compared with Control Group patients (1.09).
Compliance by the doctors with the main elements of their protocol based on the audio recordings is summarized in Table IV. Two-thirds of Control Group women received advice to stop smoking. However, more than half were also advised of two of the specific risks of smoking in pregnancy. Doctor compliance with the Experimental Protocol was relatively high. In four out of the seven main medical advice items, at least three-quarters of the Experimental Women received advice. However, for only two medical advice items was the proportion of women in the Experimental Group who received advice from the doctors significantly higher than the proportion of women in the Control Group. Doctors only informed 28% of Experimental Women of the increased risk of Sudden Infant Death Syndrome and 21% of the presence of toxic chemicals in tobacco smoke. Nonetheless, the mean doctor advice score was significantly higher (t-test, P = 0.0395) in Experimental Group patients (3.41) compared with Control Group patients (2.20).
|
Checklist data
Midwives returned checklists on 95% of Experimental Group patients while doctors achieved a 78% return rate. Midwife checklists indicated a higher proportion (81%) of Experimental Patients had been told about the six major risks of smoking in pregnancy than the proportion (23%) documented by audiotaping. In relation to the counselling section of the midwife protocol, a comparison of the midwife checklists compared with the audio recordings indicated similar proportions of women had received relevant items. For example, 76% of midwife checklists noted that midwives had discussed the counselling items pertaining to Day 1 of the Experimental Intervention compared with ratings of 57–79% of relevant audiotape items and 74% of Experimental Patients reportedly had subsequent days of the Program explained compared with ratings of 52–62% of relevant audiotape items.
Overall, doctor checklists indicated similar proportions of women had received doctor intervention items compared with the percentages based on their audio recordings. For example, the checklists noted that 82% of patients were informed of the major risks of smoking in pregnancy compared with percentages of 76–96% on the three most closely related audiotape items.
In relation to the degree to which Experimental Intervention components were delivered, the levels of statistical agreement between audiotapes and provider checklists of the same consultations is presented for four components in Table V. could not be calculated for other intervention components because the checklist items were worded more broadly than the audiotape ratings.
|
Checklist data indicate the Experimental Program was acceptable to midwives and doctors. Midwives `strongly agreed' or `agreed' that they were uncomfortable delivering their component of the Program to only 7% of patients, that they had serious difficulty securing a quit date from 9% of patients, and that their program component was a waste of time with 11% of patients. Midwives also `strongly agreed' or `agreed' that it was likely the patient would quit smoking in 21% of cases. Doctors `strongly agreed' or `agreed' that their program component was a waste of time with 6% of patients.
Provider time commitment
Table VI presents a summary of the provider time commitment for midwives and doctors. For control and experimental components, the pre-study time estimates are presented. The actual durations recorded by stopwatch are also included together with the significance of the difference between the actual durations of the Control and Experimental Components within each provider group.
|
On 96% of their checklists, midwives indicated that they `strongly agreed' or `agreed' that the Experimental Intervention took an acceptable time. After 70% of Experimental Consultations, doctors noted on their checklist that they `strongly agreed' or `agreed' that the intervention took an acceptable time.
Patient recall of smoking advice
Significantly more women (2 = 7.39, d.f. = 1, P = 0.007) had withdrawn their consent to participate in the study by the end-of-pregnancy in the Experimental Group (18%) compared with the Control Group (6%). At the midpoint, the proportions of women still participating in the trial who completed surveys were 81% in the Experimental Group and 83% in the Control Group. Table VII presents data from the midpoint questionnaire concerning the percentages of Control and Experimental Women who could recall receiving different forms of smoking advice from the midwives and doctors.
|
Patient acceptability
Of the women who completed midpoint questionnaires and who could recall receiving advice about how to stop smoking, the proportion of the Experimental Group (76%) who rated the advice as very helpful or quite helpful was significantly higher (
2 = 13.45, d.f. = 1, P = 0.0001) than the proportion of the Control Group (48%). Table VIII details ratings of general and specific features of the Experimental Program by women in the Experimental Group at the midpoint observation.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
The answer to the question `To what degree did clinic staff deliver the Experimental Program as planned?' appears to depend upon which source of information is examined. Nearly all Experimental respondents recalled watching the videotape and receiving the self-help manual. Only 3% of women in the Experimental Group could not recall any midwife or doctor advice about how to stop smoking. However, the audiotape analyses and provider-completed checklists documented significantly different levels of staff compliance with specific elements of their protocols.
Audio recordings indicated there were substantial problems with compliance by midwives and doctors. These findings are in line with the results of other studies of antenatal smoking cessation interventions where compliance by usual care providers has been reported to be a major difficulty (MacArthur et al., 1987; Shakespeare, 1990; Kendrick et al., 1995).
Midwife compliance with the risk information section of the Experimental Protocol was particularly poor, with only about one-third of women being informed of each of the six specific risks of smoking in pregnancy. The audio recordings indicated that the midwives felt the videotape and manual covered the risk information sufficiently. Given the time constraints, the decision to focus more on discussing how to quit with Experimental Women may have represented a reasonable compromise by midwives. Disturbingly, similar proportions of Control Women received the specific risk information which indicated that midwives had increased the pre-study level of usual care advice.
Compliance by midwives with the counselling section of the Experimental Protocol was higher. However, even in this section, substantial proportions of women (21–48%) missed out on being informed about specific items. Fortunately, there appeared to be little contamination of the Control Group with counselling items.
Doctor compliance, as measured by the audiotapes, was relatively high on three risk information items. However, on the three other items compliance was poor. There was also substantial contamination of the Control Group with risk information intended exclusively for the Experimental Group. On its own, however, risk information appears to have little effect on cessation rates in pregnancy (Windsor and Orleans, 1986; Walsh et al., 1997).
Comparing doctors' and midwives' compliance on the major common element of their Experimental protocols, i.e. risk information provision, the audiotapes indicated the doctors were more compliant. This was probably, in part, because the doctors' intervention tasks were less onerous. However, discussion with clinic staff suggested that the doctors may have felt more comfortable than the midwives in dealing with the biomedical issues of smoking in pregnancy.
The checklists also documented deficits in adherence to the Experimental Protocol by midwives and doctors. However, provider performance measured by checklists was higher than that indicated by the audio recordings especially for the midwives and agreement between the two measures was not significant in three out of four cases tested. These findings are in accord with the results of other studies which have demonstrated that provider-generated estimates of preventive care provision are consistently higher than those derived from record audits or consumer surveys (Lewis, 1988). Direct observation studies confirm that doctors' estimates of their rates of smoking detection and advice giving are inflated (Dickinson et al., 1989).
The compliance data support the view that a measure not dependent on provider self-report should be included to achieve more accurate assessment of provider adherence. Audio recording is logistically difficult and some providers in this study found it irksome because of its intrusive nature. Patient recall may be a more feasible option for research in antenatal clinics. Ward and Sanson-Fisher (Ward and Sanson-Fisher, 1996), although recommending the continued application of patient recall in health services evaluation of smoking care, found it was systematically biased towards over-reporting. In this study, patient recall of advice to quit was similar to that documented by the audiotape analysis. Over-reporting may not have occurred possibly because the phenomenon was attenuated by the time delay between the first visit and the midpoint measurement (Ward and Sanson-Fisher, 1996).
Pessimism about the effectiveness of smoking cessation interventions has been cited as one reason why providers fail to comply with recommendations to counsel smokers routinely (Orleans, 1985; Ockene, 1987). In this study, the midwives perception of the proportion of Experimental women who would quit (21%) was actually higher than the end-of-pregnancy rates (self-report 19% and validated 13%). Neither midwives nor doctors reported their intervention efforts were a waste of time with a substantial proportion of Experimental Women.
On average, midwives spent 28% less time with Experimental Women than requested in their protocol. In contrast, doctors devoted 44% more time than required. Therefore, it is not surprising that after almost one-third of Experimental Consultations, doctors did not agree the intervention took an acceptable length of time. Surveys have demonstrated that compared with midwives, more antenatal medical staff give a lower priority to smoking intervention (Walsh et al., 1995) and believe they do not have good skills in the area (Clasper and White, 1995). These factors may also help to explain the doctors' greater rate of dissatisfaction with the time involved.
Audio recordings indicated that much of the doctors' extra time was spent on items not contained in the medical protocol including methods of quitting. In addition, explanations of the risks involved sometimes appeared too complex. This indicates some doctors required more training in discussing the potential harms caused by smoking especially with low socio-economic class women.
Tightly controlled efficacy trials have sometimes produced very high between-group differences in quit rates both in pregnant populations (Sexton and Hebel, 1984) and in non-pregnant populations (Richmond et al., 1986). The compliance monitoring undertaken in this trial helps to explain why effectiveness can be markedly attenuated when a health education programme of demonstrated efficacy is evaluated under conditions more akin to usual practice (Kottke et al., 1990; Lichtenstein, 1997). As outlined in this trial, two phenomena occurred which were likely to reduce the intervention effect size: (1) clinic staff increased their usual level of care for Control Subjects, and (2) doctors and midwives did not comply fully with their protocols for the Experimental Intervention. This emphasizes the need for health education researchers not only to specify the content of control and experimental conditions, but also to monitor the intervention process in all study conditions.
Patient ratings of the Experimental Program indicated that it was acceptable to most women. Items relating to the doctor and midwife contributions were rated favourably by 66 and 80% of women, respectively. The Experimental Video was rated positively by 85% of respondents. This very high rating suggests videotape material may be a particularly acceptable element of an antenatal smoking cessation program. Enright et al. (Enright et al., 1988) found that audiovisual equipment was an effective means of encouraging smokers to attempt cessation. Three other Experimental Components, i.e. the self-help manual, involvement of a support person and sugarless chewing gum, were all rated favourably by a majority of respondents. In contrast, only 42% of the women strongly agreed/agreed that the lottery encouraged them to try and quit. Although assessment of the effects of quit and win smoking cessation contests in community settings has proven problematic, they probably produce only marginal increases in quit rates above the background secular trend (Chapman et al., 1993).
In primary care settings, Lichtenstein and Glasgow (Lichtenstein and Glasgow, 1992) have argued that increasing the number of treatment elements leads to better outcomes. However, Windsor et al. (Windsor et al., 1993) found that the addition of a chart reminder, medical letter and social support methods did not appear to increase quit rates in their antenatal study. It may be that similar strategies used in this trial were also ineffective.
It has been argued that pregnant women smoke cigarettes to cope with stress and undesirable social circumstances (Graham, 1976; Oakley, 1989). In a prospective investigation of unaided cessation in heavy smokers, Carey et al. (Carey et al., 1993) found that quitters perceived less stress and reported less reliance on self-criticism than non-quitters. Therefore, the extent to which cessation programs encourage guilt feelings in women is a pertinent issue which has not been discussed in reports of other pregnancy trials. Almost three-quarters of Experimental Respondents agreed/strongly agreed that the anti-smoking advice they received made them feel guilty. However, this assessment must be weighed against the findings that a higher proportion (85%) of women said they were pleased to receive advice about smoking and only 3% agreed/strongly agreed that the Experimental Program was a waste of time. Nonetheless, guilt perceptions in pregnant smokers and their significance may be an area worthy of further study.
Program ratings are only available from Experimental women who completed questionnaires. It should be noted that, by the end-of-pregnancy, three times more women in the Experimental Group had actively withdrawn from the study than in the Control Group. Clearly, for almost one-fifth of Experimental Women, participation in the trial was unacceptable. Anecdotal evidence suggests this was usually because they had decided to continue smoking and did not wish to receive advice at follow-up visits. Comparisons with other studies in relation to this point are difficult to make. For example, Windsor et al. (Windsor et al., 1993) state that 15% of their patients were lost to follow-up but no data are provided about the reasons for such losses including withdrawal of consent.
Summarizing, despite some substantial provider compliance problems, the results of the process evaluation indicate that the Experimental Program did largely meet the criteria proposed by Walsh and Redman (Walsh and Redman, 1993) for assessing whether a smoking cessation program is suitable for routine use with pregnant women. The program components were well received by most women. After a clear majority of consultations, doctors and midwives agreed the time commitment was acceptable, and that their component was not a waste of time. Training required for the program was of brief duration and was integrated into normal clinic operations. Notwithstanding the above, the process evaluation did help to identify areas where the program needed modifications.
A smoking cessation intervention based on the Experimental Program is currently being evaluated in a multi-clinic dissemination trial. Following the trial reported here, the program was, in fact, modified to increase further its suitability for routine antenatal use. To reduce the intervention complexity and the midwife time allocation, the two intervention components rated lowest by the patients, i.e. the lottery and confectionery gum, were dropped from the program. A training videotape was produced which modelled the proposed roles of medical and nursing staff, and showed how these could be performed in the suggested time allocations. In line with the findings of the compliance monitoring, the videotape recommended that only the doctors be responsible for the provision of the risk information while the midwives concentrate on the counselling component. This recommendation had the merits of further reducing the midwives' time commitment and also of reducing the likelihood of duplication of effort or conversely of each provider group assuming the other would provide the risk information. At present, only preliminary data are available on initial clinic adoption rates in the dissemination trial (Cooke et al., 1999). Full process and outcome data are not yet available.
Finally, it is important to consider some methodological limitations of this study. To reduce the time burden on the providers, some items on their checklists were worded more broadly than the equivalent items on the audiotape coding instruments. These differences in wording reduced the number of direct statistical comparisons that could be made between the two compliance measures. Future researchers need to consider carefully the comparability of the compliance instruments they employ. In relation to the audiotaping procedure, it should be noted that a non-random sampling method was used. Because consultations to be audiotaped were not randomly selected, the consultations recorded may not be representative of all the consultations in their respective intervention condition. It has been reported that direct observation measures do not appear to cause substantial reactivity problems (Redman et al., 1989). However, it is possible that the practice of audiotaping used in this study may have produced a Hawthorne effect in the consultations recorded. This could mean that the compliance problems identified by the audiotape analyses represent lower estimates of this phenomenon than actually occurred across all consultations. Patient recall data tend to support the view that reactivity may not have been a major factor, at least in relation to the providers' recommendation to quit smoking.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Process measures collected in this trial support the view that the Experimental Program for smoking cessation is suitable for incorporation into usual antenatal care procedures. However, substantial problems with provider compliance and time commitment were identified. The significant between-group differences in validated cessation rates were achieved despite these deficiencies.
Public antenatal clinics represent a particular challenge for dissemination of anti-smoking programmes because of their high workload and large numbers of staff involved. Results of the single clinic trial reported here not only demonstrated the effectiveness of the Experimental Program but also indicated ways that it could be fine-tuned to simplify its delivery so that key elements were not omitted or unnecessarily duplicated. Efforts to encourage doctors and midwives to perform more complementary roles in antenatal smoking cessation may also require extra emphasis.
The importance of teamwork and supportive structures have been highlighted in ensuring preventive programs are integrated into primary care medicine (Kottke et al., 1990). Future studies should report sufficient process measures to enable these systems issues to be examined carefully and an adequate assessment of program effectiveness to be made. Qualitative methods could be included in future evaluations to extend the heuristic value of information collected.
|
Acknowledgments |
|---|
This study was funded by a National Health and Medical Research Council Grant (PG8824). Final drafts of this manuscript were completed while R. A. W. was employed by the New South Wales Cancer Education Research Program. Thanks to Ruth Gay for assistance with rating the audiotapes, to Jayne Fryer for statistical assistance, to Karen Buxton and Anna Di Legge for typing the manuscript, and to staff and patients of the John Hunter Hospital Antenatal Clinic for their participation.
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Carey, M. P., Kalra, D. L., Carey, K. B., Halperin, S. and Richards, C. S. (1993) Stress and unaided smoking cessation: a prospective investigation. Journal of Consulting and Clinical Psychology, 61, 831–838.[Web of Science][Medline]
Chapman, S., Smith, W., Mowbray, G., Hugo, C. and Egger, G. (1993) Quit and win smoking cessation contests: how should effectiveness be evaluated? Preventive Medicine, 22, 423–432.[Web of Science][Medline]
Clasper, P. and White, M. (1995) Smoking cessation interventions in pregnancy: practice and views of midwives, GPs and obstetricians. Health Education Journal, 54, 150–162.
Cooke, M., Mattick, R. P. and Campbell, E. (1999) The dissemination of a smoking cessation program to 23 antenatal clinics: the predictors of initial program adoption by managers. Australian and New Zealand Journal of Public Health, 23, 99–103.[Web of Science][Medline]
Cummings, S. R., Coates, T. J., Richard, R. J., Hansen, B., Zahnd, E. G., Vander Martin, R., Duncan, C., Gerbert, B., Martin, A. and Stein, M. J. (1989a) Training physicians in counseling about smoking cessation. A randomized trial of the `Quit for Life' Program. Annals of Internal Medicine, 110, 640–647.
Cummings, S. R., Richard, R. J., Duncan, C. L. Hansen, B., Vander Martin, R., Gerbert, B. and Coates, T. J. (1989b). Training physicians about smoking cessation: a controlled trial in private practices. Journal of General Internal Medicine, 4, 482–489.[Web of Science][Medline]
Dickinson, J., Wiggers, J., Leeder, S. R. and Sanson-Fisher, R. W. (1989) General practitioners' detection of patients' smoking status. Medical Journal of Australia, 150, 420–426.[Web of Science][Medline]
Doherty, W. J., Becker, L. A., Burge, S. K., Whitehead, D'A. and Baird, I. (1988) Ecological analysis of an intervention study of smoking cessation in medical practices. Family Practice Research Journal, 8, 17–23.[Medline]
Dolan-Mullen, P., Ramirez, G. and Groff, J. Y. (1994) A meta-analysis of randomized trials of prenatal smoking cessation interventions. American Journal of Obstetrics and Gynecology, 171, 1328–1334.[Web of Science][Medline]
Enright, S., Kincey, J., Paxton, E. and Chapman, B. (1988) Encouraging smoking cessation. The Practitioner, 232, 846.[Web of Science][Medline]
Folsom, A. R. and Grimm, R. H. (1987) Stop smoking advice by physicians: a feasible approach. American Journal of Public Health, 77, 849–850.
Glynn, T. J. and Cullen, J. W. (1989) Health promotion through smoking prevention and cessation—the need for a selective research agenda. Health Education Research, 4, 5–11.
Gordon, J. J., Saunders, N. A. and Sanson-Fisher, R. W. (1989) Evaluating interns' performance using simulated patients in a casualty department. Medical Journal of Australia, 151, 18–21.[Web of Science][Medline]
Graham, H. (1976) Smoking in pregnancy: the attitudes of expectant mothers. Social Science and Medicine, 10, 399–405.
Haynes, R. B., Taylor, D. W. and Sackett, D. L. (eds) (1979) Compliance in Health Care. Johns Hopkins University Press, Baltimore, MD.
Hjalmarson, A. I. M., Hahn, L. and Svanberg, B. (1991) Stopping smoking in pregnancy: effect of a self-help manual in a controlled trial. British Journal of Obstetrics and Gynaecology, 98, 260–264.[Web of Science][Medline]
Kendrick, J. S., Zahniser, C., Miller, N., Salas, N., Stine, J., Gargiullo, P. M., Floyd, L., Spierto, F. W. Sexton, M., Metzger, R. W., Stockbauer, J. W., Hannon, H. and Dalmat, M. E. (1995) Integrating smoking cessation into routine public prenatal care: the Smoking Cessation in Pregnancy Project. American Journal of Public Health, 85, 217–222.
Kottke, T. E., Brekke, M. L., Solbert, L. I. and Hughes, J. R. (1989) A randomized trial to increase smoking intervention by physicians. Doctors helping smokers, Round 1. Journal of the American Medical Association, 261, 2102–2106.
Kottke, T. E., Solberg L. I., and Brekke, M. L. (1990) Beyond efficacy testing: introducing preventive cardiology into primary care. American Journal of Preventive Medicine, 6, S77–S83.
Lewis, C. E. (1988) Disease prevention and health promotion practices of primary care physicians in the United States. American Journal of Preventive Medicine, 4, S9–S16.
Li, V. C., Coates, T. J., Spielberg, L. A., Ewart, C. K., Dorfman, S. and Huster, W. J. (1984) Smoking cessation with young women in public family planning clinics: the impact of physician messages and waiting room media. Preventive Medicine, 13, 477–489.[Web of Science][Medline]
Lichtenstein, E. (1997) Behavioral research contributions and needs in cancer prevention and control: tobacco use prevention and cessation. Preventive Medicine, 26, S57–S63.[Web of Science][Medline]
Lichtenstein, E. and Glasgow, R. E. (1992) Smoking cessation: what have we learned over the past decade? Journal of Consulting and Clinical Psychology, 60, 518–527.[Web of Science][Medline]
Lumley, J. (1987) Stopping smoking. British Journal of Obstetrics and Gynaecology, 94, 289–294.[Web of Science][Medline]
MacArthur, C., Newton, J. R. and Knox, E. G. (1987) Effect of anti-smoking health education on infant size at birth: a randomised controlled trial. British Journal of Obstetrics and Gynaecology, 94, 295–300.[Web of Science][Medline]
Oakley, A. (1989) Smoking in pregnancy: smokescreen or risk factor? Towards a materalist analysis. Sociology of Health and Illness, 11, 311–335.
Ockene, J. K. (1987) Physician-delivered interventions for smoking cessation: strategies for increasing effectiveness. Preventive Medicine, 16, 723–737.[Web of Science][Medline]
Orleans, C. T. (1985) Understanding and promoting smoking cessation: overview and guidelines for physician intervention. Annual Review of Medicine, 36, 51–61.[Web of Science][Medline]
Owen, N. (1989) Behavioral intervention studies and behavioral epidemiology research to improve smoking-cessation strategies. Health Education Research, 4, 145–153.
Redman, S., Dickinson, J. A., Cockburn, J., Hennrikus, D. and Sanson-Fisher, R. W. (1989) The assessment of reactivity in direct observation studies of doctor–patient interactions. Psychology and Health, 3, 17–28.
Richmond, R. L., Austin, A. and Webster, I. W. (1986) Three year evaluation of a programme by general practitioners to help patients to stop smoking. British Medical Journal, 292, 803–806.
Russell, M. A. H., Stapleton, J. A., Jackson, P. H., Hajek, P. and Belcher, M. (1987) District programme to reduce smoking: effect of clinic supported brief intervention by general practitioners. British Medical Journal, 295, 1240–1244.
Sexton, M. and Hebel, J. (1984) A clinical trial of change in maternal smoking and its effect on birth weight. Journal of the American Medical Association, 25, 911–935.
Shakespeare, R. (1990) The development and evaluation of a smoking cessation counselling training program for midwives. In Durston, B. and Jamrozik, K. (eds), Tobacco and Health 1990. The Global War. Proceedings of the Seventh World Conference on Tobacco and Health. Health Department of Western Australia, Perth, pp. 950–951.
Thompson, R. S., Michnich, M. E., Friedlander, L., Gilson, B., Grothaus, L. C. and Storer, B. (1988) Effectiveness of smoking cessation interventions integrated into primary care practice. Medical Care, 26, 62–76.[Web of Science][Medline]
Wadland, W. C., Hughes, J. R., Secker-Walker, R. H., Bronson, D. L. and Fenwick, J. (1990) Recruitment in a primary care trial on smoking cessation. Family Medicine, 22, 201–204.[Medline]
Walsh. R., and Redman, S. (1993) Smoking cessation in pregnancy: do effective programmes exist? Health Promotion International, 8, 111–127.
Walsh, R. A., Redman, S., Brinsmead, M. W. and Arnold, B. (1995) Smoking cessation in pregnancy: a survey of the medical and nursing directors of public antenatal clinics in Australia. Australian and New Zealand Journal of Obstetrics and Gynaecology, 35, 144–150.
Walsh, R. A., Redman, S., Brinsmead, M. W., Byrne, J. M. and Melmeth, A. (1997) A smoking cessation program at a public antenatal clinic. American Journal of Public Health, 87, 1201–1204.
Ward, J. and Sanson-Fisher, R. (1996) Accuracy of patient recall of opportunistic smoking cessation advice in general practice. Tobacco Control, 5, 1–6.[Medline]
Windsor, R. A. and Orleans, C. T. (1986) Guidelines and methodological standards for smoking cessation intervention research among pregnant women: improving the science and the art. Health Education Quarterly, 13, 131–161.[Web of Science][Medline]
Windsor, R. A., Boyd, N. R. and Orleans, C. T. (1998) A meta-evaluation of smoking cessation intervention research among pregnant women: improving the science and art. Health Education Research, 13, 419–438.
Windsor, R. A., Lowe, J. B., Perkins, L. L., Smith-Yoder, D., Artz, L., Crawford, M., Amburgy, K. and Boyd, N. R. (1993) Health education for pregnant smokers: its behavioral impact and cost benefit. American Journal of Public Health, 83, 201–206.
Received on March 23, 1999; accepted on November 5, 1999
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Freund, E. Campbell, C. Paul, R. Sakrouge, P. McElduff, R. A. Walsh, J. Wiggers, J. Knight, and A. Girgis Increasing smoking cessation care provision in hospitals: A meta-analysis of intervention effect Nicotine Tob Res, June 1, 2009; 11(6): 650 - 662. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. C. Robert, J. Gittelsohn, H. M. Creed-Kanashiro, M. E. Penny, L. E. Caulfield, M. R. Narro, A. Steckler, and R. E. Black Implementation examined in a health center-delivered, educational intervention that improved infant growth in Trujillo, Peru: successes and challenges Health Educ. Res., June 1, 2007; 22(3): 318 - 331. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Winickoff, S. E. Tanski, R. C. McMillen, B. J. Hipple, J. Friebely, and E. A. Healey A National Survey of the Acceptability of Quitlines to Help Parents Quit Smoking Pediatrics, April 1, 2006; 117(4): e695 - e700. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E Campbell, R A Walsh, R Sanson-Fisher, S Burrows, and E Stojanovski A group randomised trial of two methods for disseminating a smoking cessation programme to public antenatal clinics: effects on patient outcomes. Tob. Control, April 1, 2006; 15(2): 97 - 102. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. P. Winickoff, S. E. Tanski, R. C. McMillen, J. D. Klein, N. A. Rigotti, and M. Weitzman Child Health Care Clinicians' Use of Medications to Help Parents Quit Smoking: A National Parent Survey Pediatrics, April 1, 2005; 115(4): 1013 - 1017. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. P. Winickoff, R. C. McMillen, B. C. Carroll, J. D. Klein, N. A. Rigotti, S. E. Tanski, and M. Weitzman Addressing Parental Smoking in Pediatrics and Family Practice: A National Survey of Parents Pediatrics, November 1, 2003; 112(5): 1146 - 1151. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||