Literature review
Key findings from the literature review (Stage 1) are as follows:
A. Which patients should be screened for alcohol use when pregnant or when planning a pregnancy?
Predictive factors for alcohol use during pregnancy in Australia are alcohol consumption prior to pregnancy and increasing maternal age, whereas awareness of the NHMRC Guidelines’ recommendation not to drink in pregnancy is associated with a lower likelihood of alcohol use in pregnancy [4]. As detailed in the Background, general practice and obstetric care guidelines in Australia and internationally are in broad consensus that all pregnant women should be advised of the risks of prenatal alcohol exposure and that the safest option is not to drink during pregnancy [12–20].
Timing and context of discussions around alcohol use prior to pregnancy presents challenges to clinicians. Around half of all pregnancies in Australia are unplanned, and around 55% of pregnant women report drinking alcohol prior to pregnancy recognition [11]. This suggests the importance of conversations about alcohol use for patients of reproductive age [10] and several national and international guidelines recommend this [18, 32–34]. A challenge of meeting this recommendation is that a minority of Australian women present to their GPs specifically for preconception care, unless they are accessing assisted fertility [35, 36].
B. What screening tools for alcohol use during pregnancy have been validated for use in primary care settings, both nationally and internationally?
We found ten tools that were used to screen for prenatal alcohol exposure in a variety of Australian and international health care settings, and we compared their features to determine if they were appropriate for use in Australian general practice. The features of a prenatal alcohol screening tool that make it appropriate for the Australian general practice context include that it be:
-
sufficiently brief to be used in a typical GP consultation of 10–15 minutes;
-
sufficiently sensitive to detect low or infrequent levels of alcohol use, including ‘special occasion’ drinking;
-
acceptable to primary care clinicians and patients alike.
It is also important that the tool has been validated in an Australian clinical practice context. Four of the tools identified had features suitable for the Australian general practice context. The remaining had characteristics that would render them less appropriate or practical. The tools are summarised in Table 1.
The most familiar in the Australian general practice setting is the Alcohol Use Disorder Identification Test (AUDIT) and its derivatives the AUDIT-C (AUDIT-consumption subset) and the AUDIT-C for pregnancy [15, 37–39]. Other tools that were identified as having potential value in the Australian setting are the Grog Survey App [40–42] and ASSIST [43], although none are without limitations.
Of the tools that we identified, the AUDIT-C for pregnancy [28] best fulfils the Australian general practice use criteria; notwithstanding the minor critique that the use of ‘standard drinks’ requires careful explanation to patients, and that there may be a need for an additional question to capture ‘special occasion drinking’. The AUDIT-C is speedy to administer and already widely used and validated in the Australian general practice context including in Aboriginal health settings. The differences between the tools are described in Table 1.
Table 1
Alcohol use screening tools for health care settings
| Tool | Comments |
1a | AUDIT | A 10-item screening questionnaire Alcohol Use Disorder Identification Test (AUDIT) developed by the World Health Organization. Includes questions about patterns of alcohol consumption, as well as about behavioural and attitudinal factors that suggest higher risk alcohol use and dependence. Has lower sensitivity for stop-start drinking patterns. [44] |
1b | AUDIT-C | An abbreviated form of the original AUDIT, which includes just the subset of 3 questions that enquires about the frequency and quantity of alcohol use. Widely used and validated in the Australian context and incorporated into a range of general practice clinical management systems in Australia. Validated for use in Aboriginal and Torres Strait Islander health settings. [38] When used in Aboriginal health settings, assistance often required to convert non-standard measures of alcohol intake into standard drinks. [41, 45] |
1c | AUDIT-C for pregnancy | Uses the same three item questionnaire as the AUDIT-C, but asks specifically about alcohol use in pregnancy, and has variable cut-offs for different risk thresholds to reflect that any level of alcohol consumption in pregnancy confers some risk. [28] May fail to identify women who limit their drinking during pregnancy to ‘special occasions’, suggesting that an additional supplementary question may be required. [4] |
2 | T-ACE | A 4-item alcohol screening tool, the first developed and validated specifically for use in obstetric settings (United States). Asks about alcohol tolerance rather than alcohol use with the rationale that many women may not realise that higher tolerance typically reflects higher levels of consumption and as such would feel less stigma about their response. One of the more widely used tools internationally. May have predictive value in identifying risk for neurobehavioral impacts on the child after prenatal alcohol exposure but the generalisability of this to an Australian setting is limited. [46] Not sensitive for detection of low levels or infrequent alcohol use, as it asks about indicators of dependence and tolerance while not quantifying consumption. [6, 47–50] |
3 | 4-Ps Plus | A 5-item questionnaire that aims to address under-disclosure of alcohol use in pregnancy. Developed and validated specifically for use in antenatal care settings. Asks women about their alcohol use in the month before they found they were pregnant (drinking alcohol prior to pregnancy is a strong predictor of use in pregnancy). Has high sensitivity for detection of alcohol use in pregnancy, while reducing the burden of disclosure on patients. It is quick to use. A significant disadvantage of the 4-Ps Plus tool is that it is copyrighted and requires payment of a licence to use. [51] |
4 | One Question Screen | A single question tool that may have value in the Australian primary care context: ‘When was your last drink?’ to triage pregnant patients as no risk, low risk and high risk, depending on whether the response is that they are a non-drinker, that their last drink was before pregnancy recognition, or their last drink was after pregnancy recognition. This screening tool was developed and validated in health settings in the Congo. In the event of a positive screen it requires the clinician to ask further questions to quantify alcohol consumption patterns. [52] |
5 | ASSIST | 8 item Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) tool to identify and quantify alcohol and substance use, including in pregnancy [43]. Includes consumption metrics and indicators of dependence and addiction. Estimated to take 5–10 minutes to complete. Given the duration of the average general practitioner (GP) consultation in Australia is 17 minutes, it is not realistic to complete such a screen within the already busy context of an early antenatal consultation. [53] |
6 | SMAST | A 13-item questionnaire, the Short Michigan Alcoholism Screening Test (SMAST) is designed to detect alcohol use disorders. It explores alcohol use across the lifetime. Abbreviated from the longer Michigan Alcoholism Screening Test (MAST). Used in some antenatal contexts but not developed specifically for use in pregnancy. it is [50, 54, 55]. Poor sensitivity for detecting low levels of alcohol use in pregnancy [6]. |
7 | CAGE | A 4-item questionnaire to assess perception of drinking behaviours, optimised for detecting alcohol use disorders in non-pregnant populations. It does not quantify levels or patterns of consumption and in antenatal settings it is only sensitive to detecting higher levels of alcohol consumption and dependence. [50, 55] One of the earliest alcohol screening tools developed with the intent of removing stigma around disclosure of alcohol use [56]. An adapted version called CAGE-AID is used to screen for other substance use. |
8 | TWEAK | A 5-item questionnaire that includes questions from T-ACE, MAST and CAGE (two older screening tools). [57, 58] Validated for use in pregnancy. One of the more widely used tools internationally but only likely to detect higher levels of alcohol consumption and dependence, due to its use of higher-risk criteria. [50, 54, 59] Does not quantify levels or patterns of alcohol consumption, so if positive, it requires clinicians to take a further quantitative history. |
9 | NET | A 3-item questionnaire (Normal drinker, Eye-opener, Tolerance) developed for use in antenatal contexts [60]. It has high sensitivity and specificity for high-risk drinking but is unlikely to detect low and moderate alcohol use. [61]. Has been critiqued as of high risk of “social desirability bias” due to it asking whether the person considers themselves to be a ‘normal drinker’ [50]. Given concerns around the social permissibility of occasional alcohol consumption in pregnancy in Australia, this may lead to further under-recognition of prenatal alcohol use. |
10 | Grog Survey App | A ‘patient-facing’ instrument rather than a clinician directed screening tool. Validated for use specifically in Aboriginal and Torres Strait Islander health settings. Not validated for use in pregnancy. A culturally sensitive resource that may inform the development of clinician resources. Based on a modified Finnish model, which takes a narrative approach to screening. More time consuming than the AUDIT-C, which limits its appropriateness in many general practice contexts. [40–42] |
Appropriate criteria to initiate a prompt for screening
GPs considered that entering a diagnostic code for a current pregnancy was the only feasible criteria for initiating a prompt for alcohol screening in general practice software. Clinician participants agreed that every patient known to be pregnant should be asked about alcohol use: “I think you shouldn’t make any assumptions about anything. You should just ask everybody.”
To guide discussions during pregnancy planning and preconception care, clinicians wanted the tool accessible outside of the automated prompt (visible on the clinical desktop / interface, e.g., drop down menu). GPs considered that the time constraints of general practice consultations and the rarity of patients proactively seeking preconception care, made the inclusion of an automated prompt triggered for pregnancy planning less feasible. GPs were not aware of a diagnostic code or template in general practice clinical software systems to capture ‘Planning Pregnancy’, so first the field would need to be included in the software; however, GPs highlighted that preconception advice tended to be delivered opportunistically in the context of other consultations rather than in a standalone consultation.
GPs identified a range of time points at which they might conduct opportunistic preconception care discussions (e.g. attendances for cervical screening tests and contraceptive advice), at which point they might manually access the pregnancy-related alcohol screening tool. Not all considered that a prompt would be helpful in these contexts, as they tended to judge on a case-by-case basis whether they had time to raise the discussion and to judge how an unprompted conversation about reproductive health might be received by the patient.
GPs identified that the most common timepoint they asked about alcohol use outside of pregnancy was in new patient consultations (i.e., when a patient attends their practice for the first time), and that this timepoint would be the most likely time for them to take a structured alcohol history.
Whether an automated prompt or manually accessed, clinicians were clear that they did not want any potential screening tool to be interruptive of clinical workflow; that is, they did not want to be forced to interact with the prompt to move beyond it. Passive prompts were preferred (e.g. appearing on a banner on the patient’s EMR but not requiring the clinician to interact with it).
Acceptable screening tool
All vendors reported that their clinical software systems incorporated a mechanism to document alcohol use, however, only some included formal screening tools to structure information gathering and risk assessment about alcohol use. None of the vendors were aware of having alcohol screening tools that were specific for pregnancy incorporated into their software.
AUDIT-C was the tool most incorporated into clinical software packages. Its inclusion was reported by some vendors to be relatively new and did not necessarily have prompts to use it: “(the user) basically has to go to the menu to launch it themselves”. Some vendors described other ways to capture data about alcohol, tobacco and drug use or history, e.g. “a couple of text fields and checkboxes”. One vendor described that within their pregnancy module, alcohol screening is one of several items present on a checklist for clinicians, i.e. they provided a place to document that a screen had been conducted, not a mechanism to conduct that screen.
Vendors, mostly, did not have a way to know how often alcohol screening was undertaken by software users–in particular among those presenting as pregnant: “being on premise and having a legacy architecture we don’t get a lot of data about the tools that are used… I couldn’t give you any accurate understand on how often it (AUDIT-C) is actually used”. The exception was with the cloud-based software systems where, without access to patient specific information, the vendor could generate an audit / count of how many times alcohol screening had been interacted with. One cloud-based vendor had a “consented sample” of practices from which they could, for example, check how many times the AUDIT-C assessment was used, which fields were completed and if done during antenatal or postnatal consultations (if pregnancy was recorded or if pregnancy or postnatal care was listed as a reason for visit).
AUDIT-C for pregnancy: All interviewed GPs were familiar with and/or had used the AUDIT-C screening tool, though not all had seen the adapted AUDIT-C for pregnancy [28]. Most felt that if a formal screening instrument was to be used, AUDIT-C for pregnancy was an appropriate screening tool to enquire about and quantify alcohol use (although they differed in opinion as to whether a formal screening tool would work in their own clinical consultations).
A multifunctional tool: There was, however, repeated suggestion that an antenatal clinical decision support tool should not include alcohol alone, but also encompass, at minimum, screening for nicotine and other substance use, and could include other elements of a psychosocial screen including mental health, and domestic and family violence. GPs highlighted that there is a large volume of risk relevant information that needs to be gathered in an early antenatal consult, such that it is impractical to have a separate screening tool to conduct just one facet of screening. For pragmatic reasons, GPs wanted any multifaceted screening tool to have the functionality for clinicians to partially complete, save and return to screen in a future consultation, to allow for clinicians to be able to pivot the focus of their consultations in response to what a patient discloses.
Risk stratification: Clinicians wanted the screening tool to generate scores to risk stratify patients. They emphasised the importance of the tool allowing them to assess how a patient's alcohol risk profile evolves across different timepoints. Additionally, they recommended automating the documentation of these antenatal screening scores and organizing them in a way that facilitates comparisons and tracking of changes in a patient's alcohol risk profile over time.
Repeated prompts
The prompt should repeat with the same patient over time to guard against assumption that alcohol use status pre-pregnancy or in early pregnancy has not changed.
Provide high-level management advice: Based on the risk scores, clinicians wanted the tool to provide high level advice on clinical management. There was concern about a tool providing specific recommendations that they could not feasibly action, or referral pathways that were not available in their area of practice. One GP who worked in a remote area questioned the medicolegal implications of using a screening and risk stratification tool and then not being able to follow the guidance on management, for example, if the advice was to refer to a specialised maternal alcohol and other drug (AOD) service where no such service was available.
Field auto-population: Clinicians wanted the data collected by the tools be able to auto-populate relevant fields in health summaries, electronic referrals and shared maternity care records (with patient consent) and thereby streamline workflow to avoid duplication of effort.
So if it's (screening tool) already embedded into it (clinical software) and also if it has a prompt, then it's more inviting to actually use it… if it auto-populates when we need to do a referral that again is very helpful because it just saves you time. (GP02)
Data accessible for reporting: Having reports of the captured data readily accessible to clinicians to support their quality improvement activities was suggested to increase the utility of the tool, for example, being able to assist in their demonstration of key performance indicators. GPs reported the value of this data in conducting quality improvement activities necessary to maintain accreditation, and some practices also required this to maintain funding.
Associated resources: There were concerns about the feasibility of maintaining referral resources given the constantly changing landscape of services with funding cycles, and hyperlocal referral pathways. Most GPs agreed that having one or two contacts for national or state level would be appropriate – though several GPs highlighted the lack of availability of specialist services, whether due to remote locations of practice or due to perception of long wait lists or other barriers to patient access.
Suitability for Aboriginal and Torres Strait Islander communities: A GP working in a remote Aboriginal community expressed the need to ensure that the tool is not optimised for metropolitan areas of the “southern states”, that through consultation it should reflect the needs of the people in Aboriginal communities:
I think if you were talking about rolling the software out to places that have (Aboriginal) communities, then I would do a big consultation process with Aboriginal communities. Get a council of Elders together to ask them whether they think it's appropriate to ask these questions. And of course, what you might ask in Sydney is going to be different to the Northern Territory. (GP06)
Barriers to implementation
Lack of time and incentive: The greatest barrier to using formal alcohol screening tools cited by GPs was lack of time. Vendors also raised lack of time associated with potential unwillingness to make development of a new screening tool a priority in their already full production and update release cycle schedules.
Insufficient consultation and/or projects lacking clear specifications: A barrier to development of an integrated screening tool was insufficient consultation on requirements and therefore lack of clear specifications at the outset thus requiring much “back and forth” between client and vendor–adding to the problem of lack of time.
Alert fatigue: Mentioned by both vendors and clinicians. Some vendors had policies where they actively limited the number of automated prompts (pop-ups) that a GP would receive. Vendors knew that many GPs actively disabled pop-up alerts, even when they were alerts for items requested by end-users. GPs highlighted that when they experience high volumes of prompts, that each individual alert carries less weight, and leads to desensitisation to alerts even where they may be clinically relevant. “Some people don't even look at the prompts and just push cancel, cancel, cancel.” (GP02) GPs indicated this barrier was less of a concern when prompts were non-interruptive to workflow, i.e. that they would be visible on the screen without requiring the clinician to interact with them to continue their consultation.
Technical barriers: The absence of a suitable trigger for a screening prompt before conception was considered a tricky barrier for incorporation of a pre-conception screening prompt into the software. One vendor described this as a problem that could be overcome “as long as you’ve got business logic, you understand what you’re searching for and understand what the rules are and you can translate that to the system, (then) you can do that.” Many of the barriers raised by clinician interviewees align closely with those we identified in the literature [27].
Assumptions
Some GPs acknowledged that their assumptions can be barriers to screening, with one reporting that their patients had such high health literacy that alcohol use “isn’t an issue”, but then reflecting: “Or it could be that I'm actually not picking up (alcohol use) because I'm not asking.” (GP01)
Clinical judgement and discretion: GPs indicated a range of factors they considered when making judgements about timing of discussions about alcohol use, including cues from the patient about how receptive they might be to the discussion and the perception that use of a formal screening tool will detract from doctor-patient rapport, with rapport-building potentially determining whether the patient will return.
Lack of time: Pregnant patients often booked short consultations to discuss pregnancy not realising that a long consultation is often warranted. GPs highlighted that alcohol use is one amongst many complex and often interlinked factors that contributes to pregnancy health outcomes. “The issue of alcohol is only one tip of an iceberg in how to manage a pregnancy appropriately… there’s so much information that it’s very hard to touch on the important bits and know what to deal with”.
Lack of technical know-how: The issue of the software end-users simply not knowing how to fully use the clinical software and therefore not being able to fully access its features, was raised. It was noted that many GPs do not mark a patient as pregnant within the clinical software (EMR) and so miss out on the benefit of existing clinical decision support “smarts” such as, automated advice on safe prescribing in pregnancy. Should an alcohol screening prompt for pregnant women be triggered by documenting a current pregnancy using a diagnostic code in the clinical software, then GPs who fail to use this feature would not trigger the alerts related to alcohol screening.
Structural and cultural barriers
Inadequate government funding: The lack of time to screen, reported by GPs, is in part driven by limitations in Medicare’s fee-for-service funding model for antenatal care. GPs described the negative impact of the capped value of an antenatal consultation (Medicare item number 16500) which is at a lower rate than a long consultation for a non-pregnant patient. This meant that if delivering comprehensive pregnancy care – which takes time – GPs were forced to decide between accepting a financial penalty or passing the cost of taking additional time onto patients in higher out-of-pocket fees to cover the gap. A further structural disincentive to appropriate screening is that the Medicare item number that specifically remunerates antenatal alcohol screening (Medicare item 16591) as a component of a psychosocial screen can only be claimed after 28-weeks gestation and only once per pregnancy; this does not align with clinical practice guidelines that state alcohol screening should be conducted early and often for every pregnancy. GPs reported that funding disincentives did not change their commitment to delivering high quality antenatal care, but expressed frustration that this care was devalued or made less accessible.
Under-resourced health system: GPs working in locations spanning major metropolitan settings through to remote areas of Australia indicated that they struggled to access specialist care for their patients even when clinically indicated. One GP working in an under-resourced remote community where rates of problem alcohol use were very high described a sense of futility, that gathering data about alcohol use through formal screening is little more than “just ticking boxes” unless there is the resourcing to address the underlying drivers and manage the impacts of alcohol use.
Lack of trust: GPs stressed that the approaches in working with Aboriginal and Torres Strait Islander patients must consider not just linguistic and cultural differences, but the impacts of intergenerational trauma, and the legitimate concern some community members may have that the data gathered by health care professionals may have potential to be used punitively against them, especially with regard to custodial proceedings. This was flagged as a particular barrier by a GP working in the Northern Territory, who emphasised the huge extent to which mandatory reporting laws have broken trust in the confidentiality of health information, and inhibited rates of disclosure of clinically relevant information like alcohol consumption.
Enablers of successful implementation
Vendor funding and clear specifications
Vendors wanted funding and well thought through specifications before being approached to carry out development work to integrate alcohol screening / decision support tool into general practice clinical software; yet some also wanted to be consulted on specification development to ensure “alignment and agreement from industry” and technical feasibility.
Education and training: GPs and PNs suggested key facilitators to use of the screening tool integrated into clinical software was training accompanying the release of software updates. The training should include evidence on why and how using such a tool would make a difference in terms of clinical outcomes for patients, as well as demonstrable improvement in ease of workflow in gathering, documenting, and using health data. GPs emphasised that training usually focused on clinical aspects while technical aspects of using software systems is assumed and as such, there are many clinicians within the primary care workforce with poor digital literacy.
Co-design with end-users
To ensure the tool is practical and enhances clinician workflow rather than adding to workload, with sufficient piloting and testing of the messaging to be used within the decision support tool.
Pre-consultation questionnaires
To screen for alcohol use (and potentially other substances and mental health, etc) were described as an alternative or adjunct to in-consultation screening as a strategy to maximise the value of time-constrained consultations. Some GPs reported using pre-consultation questionnaires for mental health screening in their current practice, and those who were using systems for doing this flagged the value of this in ensuring consultation time can be used as efficiently as possible. “That way we (GPs) can focus on dealing with any issues (arising from the screening) rather than trying to identify them and deal with them in the consultation.” (GP researcher) It was suggested that a ‘third party’ organisation external to the software vendor might develop the screening tool to sit outside of the clinical software, but to, ideally, integrate collated information within the clinical EMR. A barrier to conducting pre-consultation questionnaires is that it requires a mechanism for the patient to record the reason for their visit when they book, to alert the practice to send the relevant questionnaire, and few GPs reported this mechanism in place in their current practice.
Appropriate resourcing
Reform of government funding for GPs (Medicare), to ensure appropriate rebates for longer antenatal consultations, was considered a major enabler to allow time for routine alcohol screening during pregnancy. GPs stressed that funding must provide appropriate resourcing not simply to conduct the screening, but to adequately manage the care needs of the patient if they do disclose a level of alcohol use that requires further treatment. The issue of misalignment of Medicare funding and clinical practice guidelines, where the current funding model does not allow for alcohol screening early and often in pregnancy, was also raised.
Recommendation summary
Based on our literature review and stakeholder interview findings we developed a series of 11 overarching recommendations to guide the design of an antenatal screening tool and the process for its development and integration into existing clinical software systems. An additional recommendation related to reform of GP reimbursement (Medicare Benefits Scheme) to address major barriers highlighted by GPs is included. Most of the recommendations were ‘sense tested’ by stakeholders who were overwhelmingly positive. The recommendations and results of the sense testing are provided at Supplement D.