Assessment of discharge treatment prescribed to women admitted to hospital for hyperemesis gravidarum

Prescribing drug treatment for the management of hyperemesis gravidarum (HG), the most severe form of nausea and vomiting in pregnancy, remains controversial. Since most manufacturers do not recommend prescribing antiemetics during pregnancy, little is known regarding which treatments are most prevalent among pregnant patients. Here, we report for the first time, evidence of actual treatments prescribed in English hospitals.


| INTRODUC TI ON
Hyperemesis gravidarum (HG), the most severe form of nausea and vomiting in pregnancy (NVP), is characterised by persistent vomiting resulting in fluid and electrolyte derangement, weight loss, and nutritional deficiencies severe enough to require medical attention. 1 The reported occurrence of HG varies between 0.3% and 2% of pregnancies worldwide. 2 HG is the commonest cause of hospital admission in pregnancy, after preterm labour, in the UK and the United States, 3,4 and the costs associated with the burden of managing HG are significant worldwide. [5][6][7] A wide range of therapeutic agents are used in the management of HG internationally. [7][8][9] In the UK, the 2016 Royal

College of Obstetricians and Gynaecologists (RCOG) Green Top
Guidelines 10 recommend a stepwise prescribing system where the use of antihistamines and phenothiazines are defined as firstline treatment; dopamine and serotonin antagonists as secondline treatments, followed by steroids as third-line treatment; the use of pyridoxine is discouraged. Although several systematic reviews [11][12][13] provide evidence for the safety of antihistamines and dopamine antagonists used in pregnancy, manufacturers worldwide remain cautious and do not recommend their use antenatally, leaving prescribing to the discretion of the health professional when they consider the potential benefits outweigh the risks. In the light of this, it is not yet known which drug treatments are most commonly used to treat HG, or whether and how current clinical guidelines are followed by health professionals.
An understanding of the common clinical practice adopted by health professionals could help increase confidence in doctors' prescribing, highlight potential weaknesses of the actual management and therefore help improve the service offered to women affected by HG.
English Maternity Hospital Episode Statistics provide important information on HG admissions for almost all pregnancies in England, 2,14 however, they currently have no information on medications prescribed. To determine which drugs are most commonly prescribed at discharge for the management of HG in secondary care, we used anonymised clinical data from women admitted to the Nottingham University Hospitals (NUH) National Health Service Trust. We evaluated whether prescribing reflected national or local guidelines and whether prescribing differed based on women's characteristics, such as age and ethnicity.

| Study population
This study used anonymised clinical data from NUH, which is one of the largest National Health Service (NHS) Hospital Trusts in England, accounting for roughly 10 000 births a year. 15 An NHS hospital trust is a secondary care organisation, typically consisting of 1-3 collectively managed hospitals that cover a geographical catchment area for the local population. As part of England's universal Pregnancies without information on gestational week at delivery (35%) because of under-recording were excluded. This did not affect the results of our analysis as the distribution of maternal characteristics for women with known and unknown gestational age were similar (results not shown). Women admitted to hospital with a primary diagnosis of HG were identified using International Classification of Diseases version 10 (ICD-10) 16

| HG treatment medications
Prescriptions were extracted from the NUH's electronic discharge summary system, developed in-house to facilitate recording of drugs prescribed for the patient to take home, electronically issued to their general practitioner. We classified medications prescribed at hospital discharge for HG admissions according to the NUH HG management guidelines 18 as follows: antihistamines, phenothiazines, dopamine antagonists, serotonin antagonists, corticosteroids, and vitamins.
Guidelines for HG management, both nationally 10 and at NUH, 18  and gestational), asthma, and mental health problems. Women were defined with mental health problems if they had or have previously had mental health issues, such as postnatal depression, stress, eating disorders, or self-harm. Maternal characteristics were recorded during the first appointment with a midwife which is usually around week 10 of gestation.

| Analysis of drug prescribing
We plotted the occurrence of HG admissions by gestational week to assess that the distribution followed the recognised presentation of HG across pregnancy. For HG admissions with recorded drug prescriptions at discharge, we described the prescribing prevalence for each of the six drug classes at first and subsequent HG admissions.
Prescribing prevalence at first admission was assessed separately for women who only had one admission and women who went on to be readmitted. We then described the prescribing prevalence of individual drugs by admission according to whether they were first-, second-, third-, or fourth-line treatment in the NUH clinical guidelines. 18 We included separate prevalence figures to show whether drug classes (or individual drugs) were prescribed alone or in combination.
To assess whether stepwise prescribing according to the guidelines was being used universally for all women admitted with HG, we compared proportions prescribed each line of treatment across each maternal characteristic using chi-squared or Fisher's exact tests.

| RE SULTS
There were 51 613 pregnancies in the NUH pregnancy cohort between January 2010 and February 2015, of which 33 567 (65%) had a usable record of gestation. The prevalence of being admitted for HG was 1.4% in the overall population (740/51 613) and 1.7% among pregnancies with gestational week at delivery recorded (571/33 567). For those with gestational records, HG admissions peaked at 8 weeks of gestation and the median length of hospital stay was 2 days with an interquartile range of 1-3 ( Figure 1).

| Drug prescribing on discharge
There were 1037 admissions with a primary diagnosis of HG, of which 530 (51%) had records for prescribed drugs at discharge.
Antihistamines, prescribed either alone or in combination were the most common drug class prescribed at discharge from first HG admissions, followed by serotonin antagonists (Table 2).
Antihistamines prescriptions were less frequent for subsequent admissions but they were still the most commonly prescribed drug class alone. Dopamine and Serotonin antagonists were most commonly prescribed in combination at discharge from subsequent admissions. Prescribing prevalence of phenothiazines was similar TA B L E 1 Drugs typically prescribed for the management of hyperemesis gravidarum   In the context of the NUH guideline for HG management, Table 3 shows the frequency of discharge prescriptions according to each line of treatment. While promethazine was very rarely prescribed, cyclizine was generally the most common first-line drug, prescribed alone or in combination. Most first admissions had a discharge prescription for cyclizine (79% of those among women admitted only once and 83% of those for women with multiple admissions) as did subsequent admissions (9% prescribed alone and 53% in combination) ( Table 3). In terms of second-line treatments, prochlorperazine was prescribed less often than metoclopramide at discharge. The use of prochlorperazine was prescribed similarly at first and subsequent admissions, whereas discharge prescribing of metoclopramide was more common in subsequent admissions. Ondansetron, which is recommended as thirdline treatment in the NUH guidance, was in fact more common than metoclopramide for first admissions both alone and in combination with other drugs. It was also prescribed more often for discharge from subsequent admissions compared with first admissions.
The only prescribed steroid was prednisolone and, in line with the NUH guidelines, was only prescribed at discharge from readmission. Discharge prescriptions indicated that stepwise use of treatments based on HG severity was being used universally (Table 4).
First-line therapy was prescribed most often; however, third-line therapy (ondansetron) was prescribed more frequently than secondline therapy. The use of first-, second-, and subsequent-line treatments showed little variation by maternal characteristics as shown in Table 4 where the percentages are calculated over the total for each maternal characteristic value. The use of first-line therapy was slightly more common in women with recorded mental health problems. Younger women were slightly less likely to receive discharge prescriptions for second-line treatments, although there were no statistically significant differences for any other lines of treatment according to age. Lower BMI and non-smoking were associated with slightly higher use of third-line therapy. Fourth-line treatment did show variation according to certain maternal characteristics, such as deprivation and smoking status; however, it was very rarely prescribed at discharge (only 21 admissions) so statistical power was low for assessing differences. Thiamine was more commonly prescribed in women with Asian or other ethnicity recorded, and less commonly prescribed in women with previous mental health problems.

| Main findings
At discharge from HG admission, drugs were most often prescribed in combination rather than alone. The drugs most frequently pre-  Some drugs, such as steroids, are mainly given intravenously and therefore more likely to be prescribed during hospital stay rather than at discharge. Of all admissions, 49% did not have information on prescribed drugs at discharge and therefore were not included in this analysis. In UK clinical practice, women with HG are rarely discharged from hospital without antiemetics prescriptions (expert opinion from CNP and SD; SD manages NUH HG clinic admissions) therefore this missing information is attributable to suboptimal recording. A likely common reason for the lack of recording is that when women are admitted for fewer than 24 hours and discharged with the same medications as those they had at admission, the medication is not routinely re-recorded in the electronic discharge letter. We acknowledge that this exclusion could impact significantly on the results in terms of underestimation of cases who were on first-or second-line treatment previously prescribed by their general practitioner, or by the hospital if they had been admitted previously (one third of the admissions without drugs information). Moreover, the electronic discharge summary system was not universally used before 2012; therefore, some prescriptions may have not been included.

| Strengths and limitations
One third of pregnancies did not have information on gestational age, similar to national Maternity HES data, 2 and although we have no reason to think the information was not missing at random, multiple imputation of missing values could not be applied as the missing information for many variables affected the same group of women.

| Comparisons with previous studies
We found a prevalence of 1.7% for being admitted to hospital for HG   with missing gestational age) which was similar to figures from other international studies 21-23 ranging between 0.3% and 1.5%, and our previous study using national data. 24 We found a median length of hospital stay of 2 days for any admission, within the 1.8-6 day range reported in the literature. [25][26][27]  Only a few studies have so far described maternal characteristics of women by antiemetic use. In general, we found modest variation in prescribing according to women's characteristics, however, asthmatic women were more likely to be prescribed corticosteroids; it is possible that some of this use was related to asthma which is slightly more common in women presenting with HG.
Thiamine was more common in women with Black or Asian ethnicity, who are groups with particularly high risks of HG. 2 We found that older women were more likely to be prescribed phenothiazines and dopamine antagonists while women who were underweight or non-smokers were more likely to be prescribed serotonin antagonists, in contrast with two previous North American studies from 2013 30 and 2014, 31 which both reported a higher usage of antihistamines among older women. An Australian study reported that serotonin antagonists were more likely prescribed to nonsmoking women in agreement with our results. 23 The same study also found they were used more for women in higher socioeconomic groups and Caucasian women, in contrast to our study. 32 To the best of our knowledge, no study to date has assessed whether hospital prescribing for HG is compliant with national or local guidelines in England.

| CON CLUS IONS
The use of discharge medications for the clinical management of HG in a large English hospital trust showed that first-line treatment was the most commonly prescribed. However, there was more prescribing of third-line treatment (ondansetron) compared with second line (metoclopramide) showing a closer compliance with national rather than local guidelines. While there was a reticence towards promethazine and prochlorperazine treatment, steroids were confirmed to be rarely prescribed, being limited mainly to women who were readmitted.
This evidence from routine professional practice in antiemetic prescribing during gestation helps increase knowledge and confidence in implementing and delivering optimal management to women affected by HG, although we still need further information on drug prescribing for HG in primary and secondary care settings. It would be particularly useful to collect fetal and neonatal outcomes following prescriptions of different drug classes. Urgent calls are needed to fully develop and optimise comprehensive electronic prescribing systems within secondary care settings at the national level, a key data component which is currently missing for studying drug safety in pregnancy.