Hospital admissions associated with gout and their comorbidities in New Zealand and England 1999–2009

Abstract

Objectives. To describe the national demographics, comorbidities and mortality of admissions associated with gout in New Zealand (NZ) from 1999 to 2009 and compare this with English gout admission data from the same period.

Methods. The characteristics of all admissions due to or complicated by gout in NZ from 1999 to 2009 were analysed. These findings were then compared with the wider NZ population and the English National Health Service (NHS) gout admission rates from 1999 to 2009.

Results. There were 10 241 admissions due to gout (group A) and 34 318 admissions complicated by gout (group B) in NZ from 1999 to 2009. There were 32 741 admissions due to gout in England over the same period. Gout admissions rose at 5.5% per year in NZ and at 7.2% per year in England over the study period. NZ gout patients admitted to hospital were more likely to be Māori or a Pacific Islander and had 3–7 comorbidities. Multiple admissions were common with 1479 NZ gout patients admitted more than once. Comorbidities varied between the NZ groups A and B: hypertension (19–39%), renal disease (16–27%) and diabetes mellitus (20–27%) were common. Heart failure (27.6%) and cardiovascular disease (39.1%) were common in those who had gout complicating their hospital admission. This group also had poorer survival compared with those admitted primarily for gout.

Conclusion. This is the first study to describe the epidemiology of admissions associated with gout across an entire country. Admissions are rising in both countries studied and those admitted in NZ have a high rate of comorbidity and re-admission.

Introduction

Gout is a common inflammatory arthritis caused by the formation of uric acid crystals in joints. Data from the USA, the UK, Australia and New Zealand (NZ) have indicated that the prevalence of gout is increasing [1, 2].

Gout has a significant impact on the individual and society because it is associated with reduced health-related quality of life and increased use of health care resources [3, 4]. Gout also impairs function and work productivity [5, 6]. It has a significant effect on the working age population with 80% of those diagnosed with gout in a NZ primary care cohort aged 25–64 [7]. There is evidence that gout is undertreated, and when treated that there is poor adherence to therapy [8, 9].

Gout is an independent risk factor for cardiovascular and all-cause mortality and is associated with increased comorbidity [4, 10]. Hypertension, diabetes mellitus (DM), cardiovascular disease, renal disease and dyslipidaemia are all highly prevalent in gout patients [11–14]. Importantly, the number of comorbidities in gout patients is related to disease severity [12].

No study has examined admissions associated with gout across an entire country. A small study from NZ has examined 48 patients with repeated admissions for gout and found higher comorbidities and low allopurinol dosing [15]. Other inpatient studies have been small and have focused on the care of gout patients or a specific racial group [16–19]. Here we assessed hospital admissions caused by or complicated by gout over 10 years in NZ, a country with one of the highest reported prevalence rates of gout in the world (for example, ∼17% prevalence in males ≥75 years in 2008–09) [20]. As a comparison, we accessed gout admission data available from the English National Health Service (NHS) where the prevalence of gout in males ≥75 years was 2.6% in 2007 [21]. We wanted to see if their comorbidities were similar to that reported in other studies. In addition, we also assessed the change in admissions over time to see whether the increase in the population prevalence of gout was matched by an increase in gout admissions.

Accurately characterizing the epidemiology of severe gout will assist in targeting interventions to reduce its impact on health and quality of life and to reduce potentially avoidable admissions and their associated use of health care resources [12, 20].

Methods

The NZ Ministry of Health (NZMoH) collects information on all publicly funded hospital admissions in NZ. This includes health information, comorbidities and socio-economic indicators. All patients in NZ are allocated a National Health Index (NHI) identifier, which is used throughout the health system and stays with the patient for life. The NZMoH hospital admission data encompasses essentially all admissions of interest in the country, as the NZ private health system is largely restricted to outpatient consultation and elective surgery/procedures. We extracted from the NZMoH dataset all patients with a coded primary or secondary discharge diagnosis of gout [International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD10) M10 and subcodes] from a NZ hospital from 1 July 1999 to 30 June 2009. This created two groups: the first, group A, whose primary admission reason was due to gout. The second, group B, had an alternative reason for admission but had gout coded as a complicating factor. Hospital discharge coders are advised to code anything that impacts on the current admission as a complication/comorbidity (C. Lewis, NZMoH, personal communication). For each case the following were obtained: coded (anonymized) NHI, age, ICD10 primary diagnosis, ICD10 complications/comorbidities, admission date, discharge date, prioritized ethnicity, gender and length of stay (LOS). Prioritized ethnicity data assigns an individual to one ethnic group even if they have indicated mixed ethnicity. We used this type of ethnicity information because using non-prioritized data means the total ethnic population count exceeds the total actual population. We also collected and analysed the NZ Deprivation Index 2001 (NZDep01), which is a NZ measure of socio-economic status that divides the country into geographical meshblocks (median size of ∼90 people) and allocates them a rating of 1–10 depending on nine census variables reflecting material and social deprivation (1 = least deprived, 10 = most divided) [22]. For example a meshblock with a score of 1 would be in the least deprived 10% of geographical areas of NZ. Due to linking with the NZ death register, deaths occurring during the study period were also supplied by the NZMoH. The total number of publicly funded admissions across the entire period was obtained from the NZMoH. Prioritized total population ethnicity data from the 2006 NZ census was obtained from Statistics NZ. Prioritized ethnicity data from both the NZMoH and Statistics NZ used the priority order of Maori, Pacific Islanders, Asian, Other and Europeans.

Hospital practice in NZ changed over the study period and some patients seen and assessed in the emergency department were admitted to short stay units or observational wards attached to emergency departments instead of being sent straight to the ward. Individual NZ hospitals changed their reporting practices to the NZMoH for these events in a non-uniform way gradually over the study time period (C. Lewis, NZMoH, personal communication). These events are true admissions (albeit short), but most were not reported as such at the start of the time period whereas they were at the end. Therefore, for consistency across the study period we excluded all of these short stay type admissions from the temporal analyses. This meant 1250 group A and 374 group B admissions were excluded from the temporal analyses. All patients assessed in the emergency department and sent directly to the ward were included in the temporal analyses.

English (not the entire UK) gout admission data was available from the English National Health Service (NHS) [23]. Admissions for gout (ICD10 code M10) were extracted for the period 1 April 1999 to 31 March 2009. The extracted data consisted of the number of emergency gout admissions, total emergency admissions, all gout admissions and all hospital admissions.

Statistical analysis was completed using R [24]. Means were compared using t-tests, time course data were examined using linear regression. Differences between linear regressions were calculated with analysis of covariance (ANCOVA). Proportions were examined with the χ²-test. Survival was analysed between group A and group B in two ways. The first approach used the initial 5 years of admissions and assessed 5-year survival (so all examined admissions had a full 5-year follow-up). The second approach took the initial 9 years of data and analysed the 1 year survival (so again all examined admissions had a full 1-year follow-up). For both time periods survival analysis was completed with duplicated admissions firstly included and then subsequently excluded. Difference in survival was calculated with the log-rank method and median survival was calculated using the Kaplan–Meier method.

LOS values beyond 365 days were excluded from LOS calculations because extreme LOS (in some cases up to 10 years) likely reflected long-term geriatric rest home patients funded by the NZMoH (C. Lewis, NZMoH, personal communication) and also because the more frequent admissions are of greater clinical interest. This resulted in 164 group B and 6 group A LOS exclusions.

We determined the proportion of people with gout who are admitted into hospital each year for their gout in NZ. The ethnic specific population gout prevalence was determined from table 2 of a recently published NZ national gout prevalence study for the period 1 July 2008 to 30 June 2009 [20]. This study calculated population gout prevalence from administrative data based on hospitalization, drug dispensing claims, laboratory testing and primary care consultation [20]. The absolute number of group A admissions for each gender from our data set in the identical time period was used as the numerator and the ethnic specific total number with gout from ref. [20] was used as the denominator. To calculate an approximate English equivalent of the proportion of people with gout who are admitted into hospital in a year, two different prevalence figures were used because of the significant divergence reported in the prevalence of gout in England [11, 21]. An estimate of the number of English NHS admissions in the 2004 calendar year was achieved by taking 75% of the 2003–04 admission figure and 25% of the 2004–05 admission figure and adding them. This was necessary because the NHS reported gout admission figures are from 1 April to 31 March. The same procedure was repeated for the 2007 admission figures. The overall gout prevalence rate in 2004 was sourced from Anneman’s study (1.4%) and the 2007 rate was sourced from Elliott’s study (0.47%) [11, 21]. The population of England of 50.1 million (2004) and 51.1 million (2007) was sourced from the UK Office of National Statistics for calculation of the denominator [25, 26]. The study received ethical approval from the NZ Multi-Region Ethics Committee (MEC/10/019/EXP).

Results

Demographic data

Both groups A and B contained more men than women but group B cases were older by 8 years (Table 1). Women made up 20–25% of the group A cases during the study period and this did not change over the course of the study, shown in supplementary Table 1, available as supplementary data at Rheumatology Online, (P = 0.13). Woman made up 25–30% of the group B cases and this proportion decreased over the study period (P = 0.05). There was over-representation of Māori and Pacific Islanders and under-representation of Europeans and Asians in both group A and B compared with the overall NZ population (P < 0.05 all comparisons). The over representation of Māori and Pacific Islanders was greater in group A (group A vs group B: Māori 34.1% vs 24.2%, Pacific Islanders 22.2% vs 12.1%, Europeans 39.0% vs 59.0%, respectively; P < 0.05 for all ethnic group comparisons). Group A gout patients had more socio-economic deprivation compared with group B gout patients (difference in NZDep01 of 0.6, P < 0.001). Group B patients had a significantly longer LOS by 8 days (P < 0.001) and on average 3.1 more comorbidities compared with group A gout cases (P < 0.001). Group A patients were admitted under the following specialities: general medicine (36%, 3705/10 241), orthopaedics (22%, 2271/10 241), specialist internal medicine (13%, 1282/10 241), emergency medicine (12%, 1250/10 241), intensive care (5%, 509/10 241), rheumatology (3%, 303/10 241), plastic surgery (2%, 202/10 241), general surgery (2%, 197/10 241) and remainder (5%, 522/10 241).

Temporal trends

The number of group A gout admissions increased in total 61% from 1999–2000 to 2008–09 corresponding to an average increase of 5.5% per year (P < 0.001, r² = 0.94, df = 8), shown in Table 2. The proportion of total hospital admissions made up by group A gout increased 22% from 0.09% in 1999–2000 to 0.11% in 2008–09, an average increase of 2.2% per year (P < 0.01, r² = 0.76, df = 8).

The number of group B admissions was high in the first year of the study at 7300 admissions in the 1999–2000 year and then dropped sharply to a low of 2497 in the 2004–05 year. This admission rate stayed stable for the remainder of the study period.

There was a significant increase over time in gout admissions for all defined ethnic groups, shown in Fig. 1 and supplementary Table 2 (available as supplementary data at Rheumatology Online). The average yearly change in group A gout admissions for the ethnic groups was: all Europeans 4.1%; NZ Māori 5.8%; Samoan 8.3%; Cook Island Māori 12.2%; Tongan 8.3%; other Pacific Islanders 15.1%; Asian 4.6% and other/not stated −4.8%. There was no specific age group that contributed disproportionately to the increase over the study period, shown in supplementary Table 3 (available as supplementary data at Rheumatology Online).

Group B gout admissions

Cardiovascular events made up a large proportion of the primary admission diagnoses for group B gout patients with heart failure (9%), chest pain/angina (collectively 6.5%) and acute myocardial infarction (5%), all present in the top 12 admission diagnoses, shown in Table 3. Other vascular comorbidities also featured highly including stroke/cerebral infarction (collectively 3.1%) and acute renal failure (1.4%). DM was the seventh most common cause for admission at 2.5%. Infections also caused a number of admissions with cellulitis and pneumonia being common primary reasons for admission.

Repeated admissions

In group A there were 7429 people admitted on 10 241 occasions. This was made up of 5950 patients admitted once and 1479 patients admitted more than once. The total number of admissions for this repeated admission group was 4291, with a mean of 2.9 (s.d. 2.2) admissions for each individual, median of 2 and a range of 2–49 admissions. In group B there were 21 964 patients admitted on 34 318 occasions. This was made up of 15 216 patients admitted once and 6748 patients who had a repeated admission. The total number of admissions for this repeated admission group was 19 102 admissions, and an average of 2.8 (s.d. 1.6) admissions per individual, median of 2 and range 2–37 admissions.

Comorbidities

The burden of comorbidity in both group A and B gout groups was significant (Table 4). In group A gout patients 19.4% had hypertension, 14.2% cardiovascular disease and 20.2% DM. In group B patients, the prevalence of hypertension was 38.7%, cardiovascular disease 39.1% and DM 26.8%. Chronic renal disease was also very prevalent in both groups with a prevalence of 16.3% in group A patients and 26.7% in group B patients. Infection was a significant comorbidity in both groups A and B with 14.3 and 37.9%, respectively suffering from at least one type of infection.

Mortality

One-year mortality could be assessed in 8792/10 241 group A records and 31 672/34 318 group B records including repeated admissions, and 6445/7429 group A cases and 20 304/21 964 group B cases excluding duplicate admissions. Five-year mortality was assessed in 4300/10 241 group A records and 20 782/34 318 group B records including repeated admissions and 3326/7429 group A cases and 13 469/21 964 group B cases excluding repeated admissions.

Group B cases had poorer survival than group A cases both 1 year and 5 years after discharge (P < 0.001 and P < 0.001, respectively). Excluding repeated admissions and utilizing only the first admission for those admitted more than once for both groups A and B did not change the overall result (P < 0.001 and P < 0.001, respectively). The Kaplan–Meier survival plots are shown in supplementary Figs 1 and 2 (available as supplementary data at Rheumatology Online).

Median survival can only be calculated if half or more of the population die in the follow-up period and therefore this could only be calculated for group B gout patients including duplicates. Their median 5-year survival was 3.9 years (95% CI 3.8, 4.0 years). Statistics NZ has calculated the life expectancy at age 70 (the mean age in group B) in NZ for the period 2005–07 as 14.2 years for men and 16.6 years for woman [27].

English NHS admission data

There were 32 741 gout admissions between 1999 and 2009 (Table 5). There was an 86.6% increase in all gout admissions over the period, corresponding to a 7.2% annual rate of increase (P < 0.001, df = 8, r² = 0.91). There was an 86.5% increase in emergency admissions (i.e. not elective) over the period, which equated to a 16.9% annual increase (P < 0.001, df = 8, r² = 0.87). There was a significant increase in the proportion of total emergency admissions that were emergency gout admissions over the study period from 0.048 to 0.070% (P < 0.001, df = 8, r² = 0.87) and also a significant increase in the proportion of total admissions that were due to gout from 0.021 to 0.031% (P < 0.001, df = 8, r² = 0.94).

The growth in primary gout admissions between countries was not significantly different with the NZ admission rate rising at 5.5% a year and the English admission rate rising at 7.2% a year (P = 0.11). The absolute admission rate compared with the population was higher in NZ compared with England. For example, if a population of 4.25 million for NZ and 51.5 million for England is assumed, the rate of gout admission for 2008–9 was 27/100 000 (NZ) and 9/100 000 (England). If the excluded short stay admissions (see Methods section) are also included then the rise in NZ admissions over the comparable time period was significantly higher than England at 8.3% compared with 7.2%, respectively (P < 0.0001, data not shown).

Proportion of total gout population in NZ admitted per annum

The proportion of gout patients admitted due to their gout in the year period 1 July 2008 to 30 June 2009 was highest for Pacific Islanders at 2.5% (347/13 759), then Māori at 2.2% (503/22 689), European/other 0.78% (572/73 272) and Asian 0.59% (27/4598). A comparative rate for all ethnicities in England for the calendar year of 2004, assuming 1.4% gout prevalence, would have been 0.39% (2743/701 400). For the calendar year of 2007, assuming 0.46% gout prevalence, the admission rate would have been 1.6% (3796/235 060).

Discussion

This is the first report of the epidemiology of gout admissions over a decade in an entire country and it demonstrates a significant problem with gout and its management. It shows hospital admissions for gout are increasing significantly over time in both NZ and England, which is consistent with research on the rising prevalence of gout [1, 2]. Multiple studies have shown gout patients have a high burden of comorbidities and our study has also demonstrated that those admitted to hospital with gout also have a high burden of comorbidities [11–14].

The number of group B gout admissions decreased sharply in the initial period examined (1999–2000 to 2003–04) and then plateaued. The NZMoH examined this trend specifically when we brought it to their attention and could not find an explanation for it. They hypothesized that it may have been related to the education of coders. It therefore seems likely this is a coding or data collection artefact and not a true result. This represents a limitation of the group B data, and subsequently the results of this group.

There were a significant number of patients who were admitted multiple times for gout management. This suggests that these patients are suffering from treatment refractory gout, that they are not being treated adequately or that their adherence to treatment may be low or a combination of these factors. Low adherence to treatment has been documented extensively in gout [8]. Research has also found poor adherence to guidelines in the primary care and inadequate allopurinol dosing in gout patients repeatedly admitted to hospital [15, 28].

The comorbidities seen in the gout patients examined in this study have largely been seen in other cohorts [11–14]. Our rates of type 2 DM were high (20.2 and 26.8%), especially compared with the rates reported by Annemans (8.3%) and Wu (18.4% and 18.5%) [11, 12]. The rates of hypertension found in our two cohorts (19.4% and 38.7%) were lower than other cohorts, especially Phipps-Green (52–66%) and Riedel (58%) [29, 30]. Cardiovascular disease was very prevalent in our cohorts with the group B patients having a prevalence of 39.1% compared with 5.8–7.8% for the Annemans cohort and 10.2% for the Primatesta cohort [11, 14]. Our rates of renal disease (16.3% and 26.7%) were higher than Annemans primary care cohort (5–10%) but similar to the cohorts reported by Phipps-Green (30–35%) and Wu (19–30%) [12, 14]. Heart failure (7.0% and 27.6%) was low in group A but very high in group B, especially compared with the rate of 4% reported by Primatesta [14]. Overall the prevalence of gout comorbidities in our two cohorts was broadly similar to other published cohorts with some notable exceptions like heart failure [11, 12, 14, 29, 30]. The different definitions of the comorbidities, along with the differing study designs and recruitment strategies will impact on the rates reported from the cohorts.

A finding of interest is the appearance of cellulitis and pneumonia as reasons for admission in group B cases, and infection and cellulitis being common comorbidities of both groups A and B in this study. This association has not previously been reported. Infection is an important differential diagnosis of acute gout that should be considered in the diagnostic process. There are a number of reports of the co-existence of gout and septic arthritis, but this is not common [31]. The association between gout and infection could be due to infection of gouty tophi, disability, advancing age, nutritional differences, confounding or an effect of gout or hyperuricaemia on immune function. Cutaneous infections like cellulitis can mimic the skin changes of gout and vice versa, so misdiagnosis may also be an explanation. Further investigation of this interesting finding is required in large matched case–control cohorts that can provide some degree of control for confounding.

A high and increasing rate of admission, combined with a high burden of comorbidity is a significant current and future issue for western health systems like NZ and England. So what is the aetiology of this increase? There is some direct evidence linking gout and/or hyperuricaemia to hypertension, renal disease and diabetes but the sequence of causality is unclear [32–34]. The prevalence of gout and all of its major shared comorbidities are increasing [35]. This makes causality difficult to ascertain without large detailed prospective studies, augmented by experimental validation.

Strengths and limitations

There are a number of limitations of this study. The first relates to case ascertainment. Clinically gout may not be confirmed by aspiration of the affected joint. The hospital service under which the patient is admitted may influence the decision to aspirate. Other problems such as cellulitis and septic or inflammatory arthritis can mimic gout. Therefore, it is often not possible to be completely certain of the diagnosis if only administrative data are examined. Another limitation of this study relates to the accuracy of coded data collected on patients. Many factors contribute to coding inaccuracies including illegible handwriting, lack of documentation and missing files. Obesity is often clinically apparent and clinicians may not document it to avoid embarrassment due to social and pejorative implications. For obesity and dyslipidaemia it can be difficult to directly implicate them in the aetiology of a problem, and therefore they may not be included as complicating morbidities for admissions. These reasons may explain why we see such a low rate of obesity in our cohorts compared with the other NZ cohorts that have >50% prevalence (average BMI > 30 kg/m²) [30]. NZ health discharge coding of comorbidities has been shown to have moderate agreement with notes review (κ-statistics 0.40–0.74), with both under and over recording documented [36].

A major strength of the study relates to its size and inclusiveness. Having the ability to capture data for an entire country means overall trends are not skewed by regional populations or practice variation. Further research would ideally incorporate a control group, but the availability of this is limited.

In conclusion, this study shows that the rising population prevalence of gout is matched by a rising number of hospital admissions for gout, and in many cases a number of re-admissions for gout. This suggests that the amount of severe gout is increasing, and the burden on the health systems due to gout is increasing. Notably the problem, on a population level, is three times larger in NZ compared with England. This warrants increased focus on prevention of gout admissions and consequent use of health resources through improved gout management. The data are essential for identifying where interventions should be targeted at a population level and also the opportunity to target gout management in cardiovascular and diabetic patient groups.

Acknowledgements

We would like to thank Chris Lewis and Vladimir Stevanovic at the NZ Ministry of Health for supplying the gout admission data and advising on data interpretation and analysis issues. With regard to the NZ Ministry of Health data: the data is subject to Crown copyright attributed to the NZ Ministry of Health. With regard to the Statistics NZ Population data: this work includes Statistics NZ data that are licensed by Statistics NZ for re-use under the Creative Commons Attribution Non-commercial 3.0 NZ licence. With regard to the English NHS data: data supplied by NHS The Information Centre for Health and Social Care. Copyright © 2011, re-used with the permission of The Health and Social Care Information Centre. All rights reserved. With regard to the UK Office of National Statistics Data: Source: Office for National Statistics licensed under the Open Government Licence v.1.0. P.C.R. is supported by the National Health and Medical Research Council (NHMRC) Australia, Arthritis Australia and the University of Queensland Diamantina Institute.

Funding: This work was supported by the University of Otago, New Zealand.

Disclosure statement: The authors have declared no conflicts of interest.

Supplementary data

Supplementary data are available at Rheumatology Online.

References

Author notes