International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Implications of Patient Heterogeneity in Health Care Service Preferences: A Case Study in the Provision of Community Intravenous Antibiotics in the UK
David Meads, Stephane Hess, Maureen Twiddy, Carolyn Czoski-Murray, Amando Vargas-Palacios, Samantha Mason, Jane Minton

Last modified: 28 March 2017

Abstract


Background:

Currently in England and Wales, healthcare interventions are evaluated according to a medico-economic framework where only treatments which represent value for money should be provided for patients. At the same time, the NHS has pursued an agenda of increased choice for patients whose preferences might not solely be health related but cover aspects such as convenience. The provision of out-patient community intravenous antibiotics for patients with infections is an example of this. The Community IntraVenous Antibiotic Service (CIVAS) study explored patient preferences for, and cost-effectiveness of, different Outpatient Parenteral Antimicrobial Therapy (OPAT) service models. These were: receive treatment at hospital as an outpatient (HO); receive treatment at home by either a general (GN) or specialist nurse (SN); or receive training and self-administer treatment at home (SA). Each were associated with different cost, risks and benefits.

Aims:

To determine patient preferences for OPAT services, explore the heterogeneity of sensitivities across patients and explore the implications for service provision.

Methods:

We developed a Discrete Choice Experiment (DCE) based on literature review data and patient interviews to identify possible attributes and levels. Six key attributes were identified for inclusion in the DCE, along with set of attitudinal questions which would help develop the hybrid choice model. Patients were given eight hypothetical choice scenarios, each time involving the four models of care across three alternatives (with either GN or SN used for the at home treatment). The attributes (levels) were: number of treatments per day (once a day, twice a day, and pump to provide continuous infusion); appointment time given (daily appointment time given, no appointment time given, no appointment time needed[for SA]); healthcare professional administering treatment (specialist IV nurse, general nurse, doctor, self with ½ day training or self with 1 day training); communication with healthcare professional (see someone you know, see someone you don’t know;  speak on the phone with someone you know; speak on the phone with someone you don’t know); aftercare (no appointment, appointment at hospital with a nurse, appointment with your GP, telephone appointment with a nurse); risk of adverse reaction (1 in 6 chance, 1 in 10 chance, 1 in 25 chance). The attitudinal items (n=12; five-point response scale from Strongly Agree to Strongly Disagree) covered feeling towards hospitals, control over treatment and perceptions of the risks and benefits of different services and service attributes. We used a D-efficient experimental design and conducted a pilot study with 30 patients.

The data was analysed using a hybrid choice model. Attitudinal data was analysed using principal component analysis with Varimax with Kaiser normalisation to identify underlying structure within the data. Heterogeneity was explored across socio-demographics (age, gender, race, education, employment status, number of past infections), type of current infection (long/short term) as well random heterogeneity, part of which was linked to the latent attitude constructs. Patients who had experienced short or long term infections were recruited from several hospitals in England and completed the survey electronically in an interview.

 

Results:

A sample of 202 patients participated in the study (mean age=56.78, age range 20-94; 60% male; 90% white). The strongest impact on choices was found to be the type of service, followed by treatment frequency and risk of problems.

Most importantly in the context of the present paper, we found very extensive amounts of heterogeneity in sensitivities and hence preferences across patients. The socio-demographic variables we measured did affect the strength of preference for a particular service, but these effects were modest.  Much bigger differences could be attributed to random heterogeneity, both that which was linked to the latent attitude constructs and that which was purely random. For example, we found that the attitude towards the responsibility of healthcare accounted for 65% of the heterogeneity for the preference for out-patient treatment and 88% for the preference for at home treatment by a nurse. In contrast, the attitude towards hospitals accounted for only 6.9% and 4.3% for these services, respectively.

Conclusions:

While heterogeneity in preference was modest across single variables of socio-demographics, there was significant heterogeneity when multiple-strata of socio-demographics and attitudes were considered. Evaluating heterogeneity on a univariate basis or ignoring random heterogeneity and latent attitudes may thus miss important variations in preference across the same group of patients. More complex approaches to heterogeneity evaluation are required with implications for survey design and sampling. Additionally, should patient choice be a priority for the NHS, all services may have to be provided to cater for the spread of preferences; this would have far reaching financial and planning implications.


Conference registration is required in order to view papers.